Choosing a statistical test

• Choosing a statistical test depends on whether the study is

  • Test of difference or correlation
  • Test design
  • Levels of measurement
• Test of difference or Correlation

  • Association of relationship between two variables = correlation
  • Difference between 2 groups = test of difference
• Test design experimental design of the research

  • Unrelated design = independent groups is unrelated because participants are different in both conditions
  • Related design = repeated measures or matched pairs - participants are similar to each other
  • In a Repeated measures design the participants are the same they carry out both conditions
  • Matched pairs design the participants are matched based on some kind of characteristics so they are similar to each other
• Level of measurement being used means what type of data is used

  • Nominal data

  This is categorical data people are in one category or another

  • Ordinal

  Data that can be ordered in some way eg rating scales it's subjective

  • Interval

  Data that is based onto numerical scales they are objective

  Eg weight, size , speed ,scores
• Choosing a statistical test

  
• Statistical tests

  Used in psychology to determine whether a significant difference or correlation exists and consequently, whether the null hypothesis should be rejected or retained.
• Levels of measurement

  Quantitative data can be classified into types or levels of measurement such as nominal , ordinal and interval
• Chi squared

  Is a test for an association (difference or correlation) between two variables or conditions. Data should be nominal level using an unrelated independent design.
• Mann- Whitney

  A test for a significant difference between two sets of scores . Data should be ordinal level using an unrelated design (repeated measures)
• Wilcoxen

  A test for a significant difference between two sets of scores. Data should be at least ordinal level using a related design ( repeated measures) or matched pairs
• Spearman's rho

  A test for correlation when data is at least ordinal level
• Pearson's r

  A parametric test for correlation when data is at interval level
• Related t test

  A parametric test for difference between two sets of scores. Data must be interval with a related design ie repeated measures or matched pairs
• Unrelated t test

  A parametric test for difference between two sets of scores . Data must be interval with an unrelated design ie independent groups.