Chapter 7: Sampling Techniques

Sample

A subset of the population of interest

Population

entire group of interest

Census

A method of collecting data from each member of a population

Why Sample?

1. Can’t always sample everyone

2. Don’t need to study everyone (SOUP)

Sampling

Process by which researched choose a sample from a population (Focus on the HOW, not the who)

Goals of Sampling

1. External Validity

2. Representative of population

3. Economic sample/ reasonable cost (time, money, effort)

A representative sample has

external validity

external validity

generalizability

Types of Sampling Methods

Representative/Probability Sampling

1. Simple Random Sampling

2. Systematic Sampling

3. Cluster Sampling

4. Multistage Sampling

5. Stratified Sampling

Non-representative/Non-probability Sampling

1. Covenience Sampling

2. Purposive Sampling

3. Quota Sampling

4. Snowball Sampling

Representative sample is not

Biased

Representative sample

Not Biased

all members of a population have an equal chance of being included in the sample

(probability sample)

Sampling Error

The extent to which a sample is NOT representative

Unrepresentative Sample

Biased

A sample in which some members are left out

• some over-represented

• some under-represented

(non-probability Sample)

2 ways of Bias in samples

1. Convenience Sampling

2. Volunteer bias/Self Selection

Probability sampling

Random, results in unbiased/representative samples

5 Types:

1. Simple Random

2. Systematic

3. Cluster

4. Multistage

5. Stratified

What type of sampling methods are best for external validity?

Probability Sampling Methods

Simple Random Sampling

• Probability sampling

• Most basic sampling method

• Choose a random set of people from a population using a sampling frame

  • Typically use a random number generator/table

Sampling Frame

The name and contact information for all members of a population

(necessary for simple random sampling)

Systematic Sampling

• Probability sampling

• Random starting point and then choose every so many individuals

• Benefit: no sampling frame

Cluster Sampling

• Probability sampling

• Researchers divide population into natural groups/clusters and then randomly select a subset of clusters and collect data everyone in chosen cluster

Multistage Sampling

• Probability sampling

• Researchers divide population into natural groups/clusters and then randomly select a subset of clusters and collect data from a random sample within a chosen cluster

Stratified Sampling

• probability sampling

• Divide population into strata and then take data from random sample within each strata in proportion to population representation

The sample sizes of strata are proportional to population

Strata

subgroups based on shared characteristics

Majority of research uses ______ sampling

Non-probability Sampling

Non-Probability Sampling

Not Random, introduces bias, non representative

4 types

1. Convenience

2. Purposive

3. Quota

4. Snowball

Convenience Sampling

• non-probability sampling

• Sample consisting of people who are easy to contacts and readily available

• Most Common Method (quick and cheap)

Purposive Sampling

• Non-probability sampling

• Only recruiting certain kinds of people in a non-random way

ex: Want to study smoking ——> Recruit by posting a flyer at a smoke store

Quota Sampling

• Non-probability sampling

• Researchers ensure a certain number or % of people from a particular group selected non randomly (through convenience or purposive sampling)

Quota sampling vs Stratified Sampling

Quota sampling : proportional sample derived non-randomly (convenince or purposive)

Stratified Sampling: Proportional sample derived randomly

Snowball Sampling

• Non-probability sampling

• Participants recommend acquaintances/ participants ask other participants

ex: “tell your friends…parents to do this study”

Threats to representativeness

1. Self-selection

2. Sample contains too many unusual people

3. Misgeneralization

Self-Selection

People volunteer to participate

(automatic in non-probability samples, but can occur in probability ones as well)

Misgeneralization

attempt to generalize based on an unrepresentative sample

Most current studies are on _______ societies which hurts their _______

WEIRD, generalizability

WEIRD

Western, Educated, Industrialized, Rich, Democratic

12 % world population; 96% samples

How much of world population is WEIRD?

12%

How much of samples are WEIRD

96%

WEIRD samples are ________

over-represented

External validity requires _______

probability sampling

True/False: Findings can still be valid w/o probability sampling?

TRUE

Sample size is a _________ validity issue, not an ________ validity issue

statistical, external

The external validity of a sample depends on ______ not ____

How, How many

True/False: A larger sample size always makes a sample more representative

False: The representativeness of a sample relies on its composition. However, increasing the sample size tends to help increase representativeness. It’s just not a garantee

Ex:

Population: 4,000 KSU stem students

Sample 1: 1,000 (all Math majors)

Sample 2: 800 (Math, Engineering, and Biology students)