Stratified Sampling

1
Stratified Sampling Simple random sampling means
that each record in the population has the same
chance of being selected. Methods to select a
random sample include using a spreadsheet to
generate random numbers, pulling numbers out of a
hat, throwing a coin or die etc One problem is
that you might get a sample that does not really
reflect the population eg not enough girls
etc Stratified sampling ensures that the ratio of
important categories in the sample remains
similar to that in the population.
Population
Sample
2
Stratified Sampling (Part1) Eg The Exams Board
would like a sample from four schools of maths
coursework. They would like a sample of 50
pupils. Look at the table below. How many
pupils should each school provide?
Answer Total Pupils 31512395217750 A 315 x
50 21 pupils 750 B 123 x 50 8.2 ie 8
pupils 750 C 95 x 50 6.33 ie 6 pupils
750 D 217 x 50 14.5 ie 15 pupils 750
School Pupils
A 315
B 123
C 95
D 217
For a random stratified sample 21 pupils should
be picked at random from school A, 8 from B etc
3
Stratified Sampling (Part2) Eg The Exams Board
would now like a 50 strong sample from 4 schools
thats representative of the male/female
breakdown. How many males/females should each
school provide? (Use the school sample from Part
1)
Answer Total Pupils 750 A 21 150 x 21 10
females 11 males 315 B 8 61 x 8 4
females 4 males 123 C 6 60 x 6 4
females 2 males 95 D 15 122 x 15 8
females 7 males 217
School F M Pupils
A 150 165 315
B 61 62 123
C 60 35 95
D 122 95 217
For a random stratified sample 10 female 11
male pupils should be picked at random from
school A, etc