Choosing a statistical test

• Why do we use statistical testing?

  To determine if data occurred by chance
• What does statistical testing help us assess about differences between samples?

  If the difference is statistically significant
• What are the three decisions to consider when choosing a statistical test?

  1. Difference or correlation?
  2. Experimental design?
  3. Levels of data
• What does the first decision, "Difference or correlation?", relate to?

  The aim or purpose of the investigation
• What are the types of experimental design?

  Independent groups, repeated measures, matched pairs
• What is independent design?

  Different participants in each condition
• What is repeated design?

  Same participants in all conditions
• What is matched design?

  Participants matched on key variables
• What is nominal data?

  Data in specific categories, like aggressive
• How can nominal data be presented visually?

  In bar charts where bars do not touch
• What is ordinal data?

  Continuous data that can be ranked
• How can ordinal data be presented visually?

  In histograms where bars touch

  
• What is interval data?

  Continuous data that cannot be manipulated
• What level of data is "Sedimentary, metamorphic, igneous"?

  Nominal
• What level of data is "People sorted from tallest to shortest"?

  Ordinal
• What level of data is "Hair color"?

  Nominal
• What level of data is "Yes, no, maybe"?

  Nominal
• What level of data is "Temperature in Celsius"?

  Interval
• What level of data is "IQ scores"?

  Interval
• What level of data is "Very often, often, rarely, never"?

  Ordinal
• What level of data is "Dates on a calendar"?

  Interval
• What are the statistical tests mentioned?

  • Sign test
  • Chi Squared
  • Wilcoxon
  • Mann Whitney
  • Spearman’s Rho
  • Related T
  • Unrelated T
  • Pearson's
• What mnemonic helps remember the statistical tests?

  Simon Cowell Wants More Singers Receiving Unanimous Praise
• What are parametric tests?

  Tests that are more powerful and robust
• What are the three criteria for using a parametric test?

  Interval data, normal distribution, homogeneity of variance
• What is the significance of homogeneity of variance?

  Scores should have similar dispersion
• What is the difference between related and unrelated designs?

  • Related design: Same participants across conditions
  • Unrelated design: Different participants across conditions
• What is the difference between nominal, ordinal, and interval data?

  • Nominal: Categories without order
  • Ordinal: Ranked categories
  • Interval: Continuous data with fixed units
• What does the sign test assess?

  A difference in scores between related items
• What does Spearman’s Rho test for?

  A correlation when data is at least ordinal
• What does Pearson’s r test for?

  A parametric correlation at interval level
• What does the related t-test assess?

  A difference between scores in related design
• What does the unrelated t-test assess?

  A difference between scores in unrelated design
• What does the chi-squared test assess?

  An association between two variables
• What is the minimum data level for the chi-squared test?

  Nominal level
• What is the minimum data level for the Wilcoxon test?

  At least ordinal level
• What is the minimum data level for the Mann Whitney test?

  At least ordinal level
• What is the minimum data level for the related t-test?

  Interval level
• What is the minimum data level for the unrelated t-test?

  Interval level
• What is the minimum data level for Pearson's r?

  Interval level