Week 4 - Experimental design

all the individuals units of interest.

subset of units taken from the population.

sampling error is the chance difference between an estimate and the population parameter being estimated caused by sampling.

Ideally estimate is accurate or unbiased Ideally precise

each member of the population has an equal and independent chance of being selected. Minimised bias and makes it possible to measure the amount of sampling error

Create a list of every unit in the population and give each a number Decide on the number of units to be sampled (n) Use a random number generator Sample the units whose numbers match those produced by the generator

a collection of individuals that are easily available. Volunteer bias

unmeasured variable that changes in tandem with the measured variables and makes a false relationship appear. Correlation does not equal causation Gender might be a confounding variable if not accounted and controlled for in experimental design

bias in a measurement produced by unintended consequences of experimental procedures.

experimental study where two or more treatments are applied to humans.

Control group Randomisation - random assignment of treatments (pseudo-randomisation causes bias almost always) Blinding - single and double

Replication (repeat with multiple participants) Balance - all treatments have equal sample size (might result in very small overall group) Consistency of conditions Sub sample - multiple leaves on a single plant and averaged as one data point Blocking (grouping of experimental units that share the same location or time of measurement or other properties. Within each block, treatments are randomly assigned to experimental units) Randomised block design - paired design but for more than two treatments Paired design - treatments are applied to same sampling unit (two types of fertiliser tested in the same area of the field)

individual or group of individuals.

Develop a clear research question List possible outcomes Develop an experimental plan Keep the design as simple as possible Check for common design problems Is the sample size large enough Discuss the design with other people

Start with an extreme treatment to see if an unknown variable has any effect Then do a realistic experiment with lower levels

single treatment variable whose effects are of interest to the researcher.

all treatment combinations of two or more variables. Can measure interactions between treatment variables Interaction between two or more explanatory variables means that the effect of one variable depends upon the state of the other variable. Effect of gender and effect of medicine and is there an interaction between these two things?

every individual in the treatment group is paired with a control individual that has similar values for the suspected confounding variables.