Inferential tests

• Which inferential test is best for a study will depend on these criteria:

  • Whether it is a test of difference or correlation
  • Whether the data is nominal, ordinal or interval
  • Whether the experimental design is related or unrelated
• Nominal data includes tallies of discrete categories
• Ordinal data includes whole numbers that can be ordered, but are not necessarily precise measurements
• Interval data includes standardised units of measurement that can include a decimal point
• An unrelated experimental design is independent groups
• Repeated measures or matched pairs design are examples of related designs
• Chi-squared is used for unrelated tests of difference and correlations using nominal data
• The sign test is used for related tests of difference using nominal data
• Spearman's rho is used for tests of correlation with ordinal data
• Mann-Whitney is used for unrelated tests of difference with ordinal data
• Wilcoxon is used for related tests of difference with ordinal data
• Pearson's r is used for tests of correlation using interval data
• Unrelated t test is used for unrelated tests of difference with interval data
• Related t test is used for related tests of difference using interval data
• The sign test is a way to calculate the statistical significance of differences between related pairs of nominal data. If the observed value is equal to or less than the critical value, the results are statistically significant
• The null hypothesis is when a researcher proposes that the IV will not have an effect on the DV
• The alternative hypothesis is when a researcher proposes that the IV will have an effect on the DV
• To work out the calculated value for the sign test:

  • Calculate the difference between the second and first set of results
  • Add a + or - according to whether each difference is positive or negative
  • Ignore the number of equal values
  • The lower of the two scores (number of +/- signs) gives the S value
  • Compare this to the critical value- if the observed value is lower than the critical value, the results are statistically significant, meaning we can accept the alternative hypothesis