Week 3

Question 1

Problems with replication

Answer 1

Two locations likely to differ in several things, not just presence/absence of
whelks

Even though there are now three counts, the experimental treatments have
not been replicated

The experimental treatments have now been replicated
The three counts in each plot are valid sub-samples

Question 2

Problems with replication card 2

Pseudoreplication in space

Answer 2

Aim
Compare %N in leaves of different
species
Should
Sample leaves from at least two
trees of each species

Question 3

Problems with replication card 3

Answer 3

Pseureplication in time
Aim
Compare numbers in Wet and Dry
Should
Sample Wet and Dry in at least two
years

Question 4

Problems with confounding

Answer 4

Placebo effect
Experiment testing effect of Vitamin C on duration of colds
Over 50% of participants guessed correctly which group they were in
Those showed a “placebo” effect

Question 5

Problems with independence

Answer 5

Experiment objective
Test effects of “stuff” on physiology
of fish, measured by swimming,
breathing, etc.
Procedure
Catch fish and assign them to the
two experimental treatments in turn

Question 6

Problems with independence card 2

Answer 6

Experiment objective
Test effects of “stuff” on physiology
of fish, measured by swimming,
breathing, etc.
Procedure
Catch fish and assign them to the
two experimental treatments in turn
Problem
Fish will differ before experiment
starts
Solution
Assign fish randomly

Question 7

Problems with independence card 3

Answer 7

1. Study looking at plants and animals in rock pools: randomly select pools at the start
2. Returning to the same pools on later times is likely to give non-independent results
3. Study looking at plants and animals in rock pools: randomly select pools at the start
4. One solution - if possible and appropriate - is to randomly select new pools

Question 8

Enhancements to design

Answer 8

Use a balanced design when possible
Balanced designs have equal numbers of replicates in all treatments
Analysis is usually easier
The effects of violations of assumptions are usually minimised
May conflict with the requirements of a stratified sampling scheme

Use a multifactor design when possible
These designs are usually more efficient (more powerful for same effort)
These designs are usually more informative
Ensure that all combinations of treatments are included
This is referred to as an orthogonal design
Non-orthogonal designs can be difficult to analyse

Question 9

Enhancements to design card 2

Answer 9

Hypothesis (HA)
Salinity and temperature affect
survival of Trochus larvae
Experiment
Measure survival of larvae in
different combinations of salinity
and temperature

Balanced
All “cells” have the same number of
replicates
Complete
All combinations of factors – salinity
and temperature – are included

Question 10

When designing a study, with replication, what do you need?

Answer 10

e. g. whelks study
- you need more than at least two areas with the predators, and two areas without the predators
- the areas need to be spread out
- you can use sub samples so that it’s easier to count the number of whelks

Question 11

When is replication likely to be wrong?

Answer 11

There isn’t a wrong type of replication, more often it’s not enough replication that’s the problem

Question 12

Placebo effect in human studies

Answer 12

In human studies we have to worry and be very careful on placebo type effects

Question 13

In a fish catching study, why should the ease of which fish will be caught be mixed up?

Answer 13

They should be mixed up so that the fish that are caught are a suitably random selection of the population.