statistical testing

• In psychological research, it is preferable to use a parametric test as they are more powerful than non-parametric tests, but the data must meet certain assumptions before use
• Parametric tests require data to meet certain assumptions:

  • Data should be interval data
  • Data should be drawn from an underlying normal distribution
  • There should be homogeneity of variance
• Homogeneity of variance can be tested by comparing standard deviation scores for each condition
• Parametric tests of difference required are the related t-test and unrelated t-test
• If data does not meet requirements for a parametric test, Mann-Whitney U test or Wilcoxon test should be selected as an alternative
• Parametric test of correlation required is Pearson's r
• Sign test is used for paired data in repeated measures design

  • Each pair of data is scored with + or -
  • Value of S is calculated by adding total number of pluses and minuses
  • Value of N is calculated as total number of scores minus any nil scores 'O'
• For the sign test, the calculated value must be equal to or less than the critical value for the result to be significant
• Chi-Squared test can be used as a test of difference or association on nominal, unrelated data

  • Uses frequency count in a contingency table
  • Degrees of freedom calculated as df = (rows - 1) × (columns - 1)
• For Chi-Squared test to be significant, calculated value must be equal to or greater than the critical value
• Wilcoxon test is used for ordinal data or interval data not meeting parametric test requirements in repeated measures or matched pairs design
• Mann-Whitney U test is used on unrelated, ordinal data from an independent samples design

  • Calculated value must be equal to or less than the critical value for significance
• Related t-test considers related data from repeated measures or matched pairs design

  • Looks for statistically significant difference between groups using interval data
• Unrelated t-test considers data from independent measures

  • Comparison against the same table of critical values as related t-test
  • Calculated value must be equal to or greater than the critical value for significance
• Spearman's rho is used on ordinal data for correlation

  • Calculated value must be equal to or higher than the critical value for significance
• Pearson's r is used for correlation on interval data

  • Calculated value must be equal to or higher than the critical value for significance