4 Variables
At an early stage in your design you should consider what responses will be measured in the experiment.
Often the choice is obvious. The response being measured must capture biological effects in a meaningful way (maximising signal to noise), while causing the minimal amount of suffering to the subject.
There are both statistical and ethical considerations to making a choice of response variable.
Continuous variables contain more information and may require fewer subjects to measure a meaningful response, but they often take longer and more resources to measure.
Discrete variables are often simpler to measure but contain less information and so tend to require a larger number of subjects per experimental group to achieve a similarly meaningful response.
4.1 Continuous variables
Continuous variables are measured on a continuous numerical scale. In principle the response being measured can take any value in a given range.
- body weight
- survival time
- concentration of a drug in urine
- organ weight
4.2 Discrete variables
Discrete variables are numeric but can only be measured at certain fixed values. These values have direction (i.e. an increasing/decreasing scale), and the interval size between adjacent levels is implicitly the same.
- litter size
- counts of events in a time window
- clinical scores
4.3 Ordinal variables
Ordinal variables are non-numeric values on an increasing or decreasing scale, that can only be measured at set values. It is not implied that moving between adjacent levels involves the same interval size (i.e. โsmall to mediumโ may be a larger jump than โmedium to largeโ).
- disease state (mild/moderate/severe)
- pain scale
- frequency scale (never/rarely/sometimes/often/always)
4.4 Nominal variables
Nominal variables are non-numerical values, but are without any implicit ordering.
- excitability scale (fine/nervous/excited/uncontrollable)
- genotype
- sex
4.5 Binary variables
Binary variables have only two permissible values.
- paw withdrawn from hotplate (yes/no)
- present/absent
- disease state (yes/no)