Chapter 3 - Sampling

Question 1

What is a

1. population?
2. sample?
3. sampling frame?
4. representative sample?
5. sample bias?
6. probality sample
7. sampling error

Answer 1

1. universe of units from which the sample is selected
2. segment of population saelected for investigation
3. list of all units
4. a sample relfecting population accurately
5. distortion in representativeness
6. sample selected using random selection
7. difference between sample and population

Question 2

What is a sampling error? In which fields might these errors occur? how can these errors be reduced?

1.
2.
3.
4.

Answer 2

Difference between sample and population

• biased sample: not representing population, over or under- representation of groups
• sources of bias: non-probability bias, inadequate sample frame, non-response
• can be reduced by probability sampling

Question 3

What are the types of probability samples?

1.
2.
3.
4.

Answer 3

SY SI ST MU
- systematic sample (selects unit direclty from sampling frame, e,g. every 4th unit, make sure it has no inherent ordering and if yes, rearrange)

• simple radnom sample (each unit with equal probability of selection : n/N, using random numbers)
• stratified random sample (categorise population into strata, then sample representative of each stratum, then random select within each category)
• multi stage cluster sample (useful for widely dispresed populations, first divide in clusters, e.g. industries, then sample sub clusters from these, then random select units, then collect data of each unit)

Question 4

A probability sample should be….

1.
2.
3.
4.

Answer 4

• representative
• pass an inferential statistical test
• 95% of samples should be between +/- 1.96 Standard error from population mean -> confidence interval
• standard error: SE - estimate of discrepancy between sample mean and population mean

Question 5

The larger the size we caclulate the SE from…..

Answer 5

….. the more precise and representative it is likely to be

Question 6

What factors affect a sample size?

1.
2.
3.
4.

Answer 6

T H A N

• time and cost ( after certain point a bigger sample does not bring more certainty)
• non-response (response rate is the percentage of sample who agreed to participate)
• heterogenity of population (variation of population enlarges sample size)
• kind of analysis to be carried out, some techniques require large samples, e.g. inferntial statistics

Question 7

Which types of non.-probablility sampling are there?

1.
2.
3.

Answer 7

Q S C

1. convenience sampling (most easily accessible individuals, useful when trying out new instrument)
2. snowball sampling (initial contact wioth small group, respondents introduce others in their network
3. Quota sampling ( often used in market researh, proportionately representative of populations cosial categoreis (strata)