sampling and randomisation

Question 1

population

Answer 1

parameters (unknown and hard to measure); the large group of units about which inferences are made eg. all married women
mew

Question 2

sample

Answer 2

statistics - the smaller group of units that are measured
x bar
used to make inferences if representative

Question 3

advantages of sample survey over a census

Answer 3

census is not always possible
speedier
less costly
can devote more resources to getting an accurate sample

Question 4

sampling data

Answer 4

define population of concern
specific the sampling frame - electron roll, telephone list etc. (representative)
specify sampling method - haphazard and convenience or probably and randomness
determine sample size to achieve desired accuracy - error is alpha 1/squrt(n)
sample data
apply statistical description and inference

Question 5

types of sampling

Answer 5

• convinience and haphazard
• random (probability sample)
• stratified random (probability sample)

Question 6

convenience and haphazard

Answer 6

selection bias: readers of column where survey is are only a subset of population

voluntary response bias: people with opinions do surveys

non-response bias: people don’t respond or can’t be contacted

undercoverage bias: small sample, does not represent population

questionnaire bias: bias can be introduced on how something is worded

Question 7

simple random sampling

Answer 7

all units in population have the same chance of being in the sample
usually representative if the population is homogeneous
can do randomisation in excel

Question 8

stratified random sampling

Answer 8

sort into groups (age, sex, some confounding variable) and then randomly allocate (proportional to stratum in population)

• cluster sampling - sample an entire group
• systematic sampling - randomise list and select a random starting point and choose every tenth unit
• multistage - combine methods
• helps with a heterogenous population