Principles of Sampling (I): Design

Question 1

What is Sampling?

Answer 1

Taking data for a sub-set of the population

Question 2

Why do we Sample?

Answer 2

Not possible to enumerate the whole population
(Resources, time, logistics and cost)
Therefore…
- Want to provide estimates that are precise and accurate
- Limit bias

Question 3

What is inferential/statistical testing?

Answer 3

Estimation of the population parameters using sample statistics and sample analysis
If the sample is big enough this should reflect the true estimates of population parameters

Question 4

What are the 5 types of sampling?

Answer 4

Simple Random Point Sample
Systematic Point Sample
Areally Stratified Point Sample
2-Stage area cluster point sample
Stratified, Systematic and unaligned point sample

Question 5

What is a ‘Simple Random Point Sample’?

Answer 5

Randomly selected XY coordinates

Can lead to clustering, and uneven coverage

Question 6

What is ‘Systematic Point Sampling’?

Answer 6

First point is randomly selected
Following points are at regular, pre-determined intervals
May miss variability if there’s an underlying pattern

Question 7

What is an Areally Stratified Point Sample?

Answer 7

Area divided into cells and strata
Points randomly selected in each cell
More evenly sampled and distributed
More precise

Question 8

What is the 2-Stage area cluster point sample?

Answer 8

Area divided into cells and strata
Cells and Strata randomly selected
Random points within cells
Good for large areas
Could miss local specifity

Question 9

What is the Stratified, Systematic and unaligned point sample?

Answer 9

Combination of previous designs
Divided into Strata
Sample row numbers for X coordinates
Sample column numbers for Y coordinates
Most accurate and precise

Question 10

What is the Hypothetical Small numbers problem?

Answer 10

Small absolute numbers produce highly uncertain and unreliable ratios and percentages

Question 11

What is the Small Numbers Problem?

Answer 11

Small populations can show the wrong pattern

Therefore only use data that’s statistically siginificant

Question 12

What has to be considered before making inferences about populations, from samples?

Answer 12

Is our sample a ‘good’ estimate of the population? In the long run? On average?

Question 13

How can we see whether our sample is representative?

Answer 13

Mathematical Analysis

Simulation (repeat many times to see what happens)