# HSCI 307 Lecture Notes - Lecture 9: Stratified Sampling, Cluster Sampling, Sampling Probability

56 views2 pages

Published on 24 Jan 2018

School

Department

Course

Professor

Non-probability sampling

•

Probability sampling

•

…depends on primarily on whether reliable inferences are to be made about the population

•

Sampling is a means of selecting a subset of units from a population for the purpose of collecting

information from those units to draw inferences about the population as a whole

Uses a subjective method of selecting individuals or units from a population

•

Fast, easy, and inexpensive

•

Non-probability sampling

Selection based on the principle of randomization or chance

•

Reliable estimates can be produced along with estimates of the sampling error, and

inferences can be made about the population

○

More complex, time-consuming, and usually more costly than non-probability sampling

•

Probability sampling

Non-probability sampling - Non-random samples

Unclear whether sample is generalizable

○

Sample participants selected based on their relevance to the research topic rather than

their representativeness

•

The selection of individuals from the population for a non-probability sample can result in

large bias

•

Selection bias - an individual's inclusion probability cannot be calculated for non-

probability samples, so there is no way of producing reliable estimates of their precision

(sampling error)

•

It is unclear whether or not it is possible to generalize the results from the sample to the

population

It can be used to generate ideas

•

As a preliminary step toward the development of a probability sample survey

•

As a follow-up step to help understand the results of a probability sample survey

•

It is often used to select individuals for focus groups and in-depth interviews: Census of

Population questionnaires

•

Too costly

○

Participants cooperation difficult with stigmatized behaviours

▪

Difficult to obtain large sample of rare groups (e.g. heroin users)

▪

Difficult to capture "hard-to-reach" populations

▪

Not methodologically viable方法上可行

○

Often probability methods are

•

Why use it?

Individuals are selected in an aimless, arbitrary manner with little or no planning

involved

○

Assumes that the population is homogeneous

○

Haphazard sampling - "convenience sampling"

•

Based on previous ideas of population composition and behaviours

○

An expert with knowledge of the population decides which individuals in the

population should be sampled - the expert purposely selected what is considered to

be a representative sample

○

Perhaps more bias than haphazard sampling

○

Can be useful in exploratory studies, e.g. in selecting members for focus groups or

in-depth interviews to test specific aspects of a questionnaire

○

Judgement sampling

•

Volunteer smokers, diabetics, people with sleep disorders…

○

Can be subject to large selection bias, but is sometimes necessary

Volunteer sampling

•

One of the most common forms of non-probability sampling

○

Sampling is done until a specific number of units for various subpopulations (the

quotas) has been selected (20 males and 20 females)

○

Is considered preferable to other non-probability sampling because it forces the

inclusion of members of different subpopulations

○

Similar to stratified sampling but differs in how the individuals are selected -

relatively inexpensive and easy to administer

○

Quota sampling

•

Probability sampling

i.

Non-probability sample, usually a quota sample

ii.

Improve quota sampling by using a combination of probability and non-probability

sampling.

○

Modified probability sampling

•

6 types of non-probability sampling schemes

Probability sampling

Random refers to a selection process that gives each

element/unit in a population an equal probability of being

selected

○

Individuals must be randomly selected

1.

Not that all units have the same inclusion probability but

that all units have a known non-zero inclusion probability

○

SRS and SYS are both equal probability sample

○

Individuals must have a non-zero chance of being selected

2.

2 main criteria for probability sampling (QUIZ)

Since each individual is randomly selected and each individual's

inclusion probability can be calculated

•

Reliable estimates of interest can be produced (e.g. prevalence,

incidence or association, and an estimate of the sampling error of

each estimate)

•

Main advantage of probability sampling

It is more difficult, takes longer, and is usually more expensive

than non-p sampling

•

Main disadvantage of probability sampling

Types of probability sample designs

every possible sample of size nhas an equal chance of

being selected

▪

Each individual in the sample has the same inclusion

probability - π = n/N, N is the number of units in the

population, (e.g. lottery 6/49)

▪

The most basic

○

If the objective of the survey is simply to provide overall

population estimates - this should be sufficient

○

Advantage: simple to conduct

○

Can be expensive

▪

Requires list prior to sampling

▪

Disadvantages

○

Simple random sampling (SRS)

•

A gap, or interval between each selection (e.g. every 5th

house in a street)

○

Tells how many elements to skip in the sampling from

before you pick of your sample

▪

The sample size n

1)

Population size N

2)

To calculate N/nd

▪

Sampling interval

○

Systematic sampling

•

Individual

If the cost of survey collection is high and the resources are

available…less statistically efficient sampling strategy than

SRS

○

The entire population is divided into clusters or groups and

a random sample of these clusters is selected 分组后随机选

组

○

Typically used when the researcher cannot get a complete

○

Cluster sampling

•

Groups/clusters

Less focus on representativeness

•

Greater focus on how the

relevance of the sample to the

research topic

•

Interest in cases that will enhance

what the researchers learn about

the processes of social life in a

specific context

•

Non-probability samples

•

Qualitative perspective

Representativeness

•

Produce accurate generalizations

about larger group

•

Saves time and $$

•

Measurement more accurate (e.g.

more effort on high-quality

measurements)

•

Probability samples

•

Quantitative perspective

Terminology

The concretely specified large group of many cases

from which a researcher draws a sample and to

which results from a sample are generalized

•

The population of interest about which inferences

are desired

•

Target population (source population?)

A selected subset of the target population

•

Sample (sample?)

A member of the target population (e.g. a person,

a health facility)

•

Sampling unit/element

A list of all available sampling units in the target

population (e.g. telephone list, electoral list, list of

schools, driver's license records)

•

Sampling frame (study population?)

The ratio of the size of the sample to the size of the

target population

•

Sampling ratio

Chapter 8 Sampling

Friday, March 24, 2017

15:35

week 9 Chapter 8 Sampling Page 1