Toolkit 5 Secondary Data Analysis And Processing

Question 1

What is a standard error

Answer 1

The difference between sample estimate and population parameter. Standard error should be a random error

Question 2

What are the aims of a good survey sample?

Answer 2

Valid, reliable, representative, unbiased conclusions about genuine relationships
Right people to say things to measure and record
Sufficient people to ask/ things to measure
Right questions / appropriate measurements

Question 3

What is probabilistic samples?

Answer 3

Known probability
Description, estimation of population characteristics, testing hypothesis
Extensive designs
Representative

Question 4

What is non probabilistic samples?

Answer 4

Theory development, testing instruments
Often used in small scale sociological work
Perhaps representative of a small sub group
Hidden populations

Question 5

Probabilistic methods

Answer 5

Simple random sampling
Systematic sampling
Multistage / clustered sampling - dividing population into parts and interviewing within some of parts, chosen randomly
Statified sampling- dividing populations into parts and interviewing appropriate fractions within all parts
Large scale surveys often combine clustering and stratification

Question 6

Advantages of a random sample

Answer 6

Known chances of each respondent being included
Guaranteed to be representative so unbiased sample
Sample precision increases with number of samples

Question 7

Disadvantages of random sample

Answer 7

Requires complete sampling frame
Other methods have more control- greater precision for given n
Big distances to respondents - travel time

Question 8

What is systematic sampling

Answer 8

Go systematically through a sampling frame eg every 100th person

Question 9

Advantages of systematic sampling

Answer 9

Known chances of each respondent being included
Guaranteed to be representative so unbiased sampling
Sample precision increases with number

Question 10

Disadvantages of systematic sampling

Answer 10

Still need whole sampling frame
Problem of bias if interval is systematically related to observations
Standard errors are not random they are systematically related to sample characteristic

Question 11

What is multistage/ clustered methods

Answer 11

Most often used in large scale extensive academic studies and in large scale government surveys
Have to find optimal clusters

Question 12

Advantages of clustering

Answer 12

Don’t need complete sampling frame at lowest level
Cheaper- don’t need to cover whole country for first visit/ posting or for call backs
Less cost and time

Question 13

Disadvantages of clustering

Answer 13

Gives false estimates of the size of the variability
People within clusters ar more alike than general population
Therfore a lower precision, increased standard errors determined by sample design

Question 14

What is stratified strata methods

Answer 14

Are internally homogenous groups. similar within but different between. Stratify according to the characteristics assumed to be closely related to the variables under study

Question 15

What are stratified method methods?

Answer 15

Define the strate and then sample from the strata in proportion to its size eg gender

Question 16

Advantages of stratifying

Answer 16

Ensures small but numerically vital categories eg ethnic minorities
More precise results for given n and hence more efficient use of resources

Question 17

Disadvantages of stratifying

Answer 17

Need prior knowledge to define the strata - demands of sampling frame grow substantially

Question 18

What is the Health Survey of England an example of?

Answer 18

A combination of stratified and multistage design

Question 19

What is snowball sampling?

Answer 19

Find one respondent and ask more names and so on eg drug users
It is often used in qualitative work

Question 20

Advantages of snowball sampling?

Answer 20

May be the only possible method
Personal recommendation may improve chances of gaining entry
Hidden populations

Question 21

Disadvantages of snowball sampling?

Answer 21

What is it really sampling

May miss solitary people/ activities

Question 22

Quota sampling

Answer 22

Often used in commercial research as it enables street interviews. The sample is made representative by imposing quota controls linked to the topic eg age, gender, marital status

Question 23

Advantages of quota sampling

Answer 23

Easy to do in the field
Interviewing costs are low
Traveling time and distances reduced
May only be one way to stratify

Question 24

Disadvantages of quota sampling

Answer 24

Can’t tell if the results are representative
May miss those who don’t go out
High risk of interviewer bias
Boundaries of quotas often under represented

Question 25

What is a simple random point sample?

Answer 25

Obtains map with x and y co-ordinates

The product is an uneven pattern with some areas over sampled and others under sampled

Question 26

What is systematic point sampling?

Answer 26

Eliminates an uneven distribution of points across study area
Overlay study area map with regular grid of desired size
Obtain desired number of points at regular intervals beginning at one point
The result is a representative and proportional coverage of study area

Question 27

What is stratified point sample?

Answer 27

Whether it is proportional or disproportional depends on problem at hand eg if area a comprises 60% of entire study area and area b comprises remaining study area, then the proportional stratified sample 60% of all the points will come from area a and 49% will come from area b

Question 28

What is cluster point sample?

Answer 28

It is cost effective but ignores large parts of study areas so it is necessary to make sure that the under representation will not comprise the study

Question 29

What is population

Answer 29

It is a theoretical construct and can be hard to define at times