Research Skills - Experimental Design

null hypothesis

states that the outcome is not different between two (or more) groups

what needs to be in a hypothesis

• what you are changing

• what you are measuring

• the model (cell, tissue, animal, human)

how to choose what model to use in an experiment

depends on stage of research programme, physiological systems, ethics, previous research

use ‘lowest’ model in first instance to build up support

i.e. we use isolated tissue then rats then rodents then primates then humans

3 types on controls

1. time matched control

2. vehicle or solvent control

3. positive control

what is a time matched control?

a control group evaluated at the exact same chronological time points as the treated or test group

so same experiment carried out on another tissue simultaneously

what is a vehicle or solvent control

a control group which does not have the drug dissolved in any solution to ensure the solution/vehicle does not impact the results

so in this control it is important to ensure the concentrations are the same

what is a positive control

it is an experimental control group that receives a treatment with a known effect

this is only carried out if no response is achieved

what is likely to cause bias?

human errors - e.g. no blinding

is sample size n=2 adequate?

no, it is too small to even carry out any statistical test and confirm results, replicates are important in providing accurate data

if data are normally distributed

calculate the group mean and standard error of mean

if data are not normally distributed

calculate the group median and interquartile range

if experiment done on same tissues collected from the same 6 rats do we use paired or unpaired t test?

paired t test

if experiment done on different tissues collected from 6 rats, even if same 6 rats do we use paired or unpaired t test?

unpaired t test

2 tailed t test means

results can increase or decrease

two directional

if we have more than 2 groups and data is normally distributed we use

ANOVA - one way analysis of variance

what does ANOVA tell us

whether there is a difference between one or more groups

what is the advantaged of presenting standardised data

avoids problems arising due to wide variation in size of readouts between different samples