Example 5

Effect of THC on body temperature

View experiment diagram | Use this diagram as a template

An experiment is conducted to test the effect of THC on body temperature in rats. The null hypothesis is that THC has no effect on body temperature in male rats and the alternative hypothesis is that it does have an effect. These are indicated in the properties of the experiment node.

To test this hypothesis, rats are randomised into three groups. The allocation strategy is a randomisation within blocks, with time of the day used as a blocking factor. This is indicated on the diagram with a nuisance variable node with two categories: morning and afternoon. Twenty four animals (as determined by power calculation) are randomised into 3 groups of eight. This is done using the allocation sequence generated by the EDA. To enable blinding, the spreadsheet is directly emailed to a colleague not directly involved in the experiment, who will prepare the 24 syringes so that the persons injecting the animals and measuring the temperature are unaware of the treatment each animal receives.

Group 1 is the control group, this is indicated in the properties of group 1. Animals in this group will receive an injection of vehicle and animals in group 2 and 3 will receive an injection of 1 mg/kg and 3 mg/kg THC, respectively. All injections are given intraperitoneally; this, along with the potential adverse effects is indicated in the properties of the pharmacological intervention nodes.

An hour later, body temperature is measured; for each animal the temperature is taken as an average over 5 minutes. The outcome measure (response variable) is continuous. The data will be transformed before analysis because it is anticipated that the variability of the response will increase with the size of the response, a log transformation should correct that and make the data amenable to parametric analysis. This is indicated on the diagram with a data transformation node.

In this experiment there is one independent variable of interest: THC, which has three categories: vehicle, 1 mg/kg and 3 mg/kg. It is included as a factor of interest in the analysis. The nuisance variable: Time of the day is included in the analysis as a blocking factor. If the data fits parametric assumptions (after log transformation), it can be analysed with a one-way ANOVA with one blocking factor (also called two way ANOVA without interaction).

The analysis is also carried out blind, the data is organised by the third party the allocation sequence was originally sent to into ‘Group A’, ‘Group B’ or 'Group C', so that the person doing the analysis knows which animals are grouped together but not what treatment the animals received.


Three-group comparison | Blocking factor | Block randomisation | One-way ANOVA | Data transformation

First published 30 April 2015
Last updated 20 September 2023