Randomised block design

• When does randomised block design need to be considered?


  When the effect of a second variable other than the one being investigated may affect results.
• What is a blocking factor?


  A blocking / nuisance factor is the name for the secondary variable that could possibly affect the results and conclusions.
• Completely randomised design - this is when the elements being tested are randomly assigned to the various experimental groups.
• Advantage of completely randomised design


  Quick and easy to sort out
• Disadvantage of completely randomised design


  There is a possibility that all of a certain gender, age, treatment end up in the same group meaning that any changes may be to the other variable.
• Relevant hypothesis tests to completely randomised design

  • One factor ANOVA
  • Unpaired z test
  • Unpaired t test
  • Wilcoxon rank sum
• Randomised block design - when the effect of the blocking factor is removed by ensuring that this factor is represented equally across the various experimental groups.
• Advantages of randomised block design


  Removes effect of another variable to ensure more reliable conclusions
• Disadvantages of randomised block design


  Take longer to set up the experiment to ensure equal representation
• Relevant hypothesis tests to randomised block design


  • Two factor ANOVA
• Matched pairs design - where elements are paired across the experimental and control group based on variables which they share e.g. gender. This is so direct comparisons can be made.
• Advantages of matched pairs design


  Eliminates individual variation - less biased.
• Disadvantages of matched pairs design 


  • may not be possible based on volunteers
  • can take a long time to set up
  • may be issues if people drop out
• Relevant hypothesis tests to the matched pairs design 


  • paired sign test
  • paired wilcoxon signed rank
  • paired t test