Inferential Testing

• 8 types of inferential statistics

  Mann Whitney, wilcoxon, chi square, sign, spearman’s, related t, unrelated t, pearson’s
• the 3D’s when choosing an inferential test

  difference? (or correlation)

  design

  data
• Level of significance

  The level at which we say results are meaningful or significant
• what level of significance is used in psychology?

  p<0.05
• why do we use p<0.05 when talking about levels of significance?

  it represents a balance between the risks of making a type I and type II error
• type I error 


  setting a probability level that is too low or lenient so we are more likely to accept the alternative hypothesis
• type II error 


  Setting a probability level that is too high so We are more likely to reject the alternative hypothesis
• critical values


  the point at which results from an inferential test become significant
• observed value 


  the value that is directly observed as a result of the experiment
• how to calculate the observed value

  • work out whether the ppt is +, - or 0
  • count number of +
  • count number of -
  • smallest one is the observed value
• how to calculate the critical value

  • work out N (number of ppts - number of 0s)
  • make sure you know the level of significance and whether the hypothesis is one or two tailed
  • if the OV<CV, the results are significant
• Sign test

  Repeated/matched design, nominal data
• wilcoxon test

  repeated/matched design, ordinal data
• Related t test

  Repeated/matched design, interval data
• Chi square test

  Independent groups design, nominal data
• Mann Whitney test

  independent groups design, ordinal data
• unrelated t test

  independent groups design, interval data
• spearman’s rho test

  correlation, ordinal data
• pearson’s test

  correlation, interval data