Eliminating Bias 2
Randomisation
Randomisation is an essential step in designing an experiment. Each experimental unit must be numbered.
For a completely randomised design use consecutive numbers.
For a randomised block design, number the experimental units within a block
You could use:
Physical Randomisation
Assume you have 24 rats to be randomised to 4 treatment groups. Write the numbers 1-24 on separate pieces of paper. Fold, put in a hat or other receptacle, shake well. Take out six pieces of paper and assign the corresponding rats to treatment 1, and so on.
Tables of Random Numbers
Most statistical textbooks have tables of random digits. Suppose you wish to randomise 30 mice to six treatment groups. Start somewhere in the middle of the table and look for consecuitive pairs of digits. Ignore any that are greater than 30. When you find a number between 01 and 30, say 16, assign that experimental unit to treatment 1. When treatment 1 is full, go on to treatment 2 etc.
Computer Programs
Some computer programs such as MINITAB will allow you to put a series of numbers (say 1-24) in a column, and then randomise them into the next column. If you want to randomise 24 guinea pigs to 4 treatment groups, then the first six random numbers should be assigned to treatment 1 etc.