choosing a statistical test

• 1. difference or association- can figure out from hypothesis
  2. experimental design- related=matched pairs or repeated measures unrelated= independent groups
  3. level of measurement- nominal=categorical and discrete data, ordinal=scale, data is ordered not have equal intervals between, and interval= based on numerical scales, equal intervals, precise, and parametric tests
• chi squared:

  can test for difference or association

  if difference the experimental design is unrelated

  nominal data as data is discrete or categorical
• mann whitney:

  tests for a difference

  unrelated experimental design

  ordinal data as data is ordered or on a scale but not in equal intervals
• unrelated t-test:

  tests for a difference

  unrelated experimental design

  interval data as based on numerical scales, equal intervals
• sign test:

  tests for difference

  related experimental design

  nominal data as discrete or categorical
• wilcoxon test:

  tests for a difference

  related experimental design

  ordinal data as data is ordered but not equal intervals
• related t-test:

  tests for a difference

  related experimental design

  interval data as data is based on numerical data and has equal intervals between
• spearmans rho:

  tests for an association or correlation

  ordinal data as data is ordered but not in equal intervals
• pearsons r:

  tests for an association or correlation

  interval data as data is based on numerical scale and has equal intervals between
• if test has an r in then calculated value needs to be higher than critical for it to be significant

  if test does not have an r in then calculated value needs to be lower than critical for it to be significant