Transcript Research Methods - Albright College

Research Methods
Chapter 5:
Sampling
Sampling


Purpose: To draw enough of something to
make your findings generalizable
Some things to do before conducting a
sample


Consider a census
Evaluate Generalizability



Sample generalizability: Can the findings from the sample
be generalized to the population from which that sample
was taken?
Cross-population generalizability: Can the findings from
one population be generalized to another slightly
different population?
Assess the diversity of the sample

Shoot for a representative sample…
Sampling Methods

Non-probability Samples


Often used in qualitative research
Probability Samples

Often used in quantitative research
Nonprobability Sampling

Why do it?



Random sampling may not be possible or is
too expensive
May be doing exploratory research
Types of nonprobability samples




Availability sample
Quota sample
Purposive sample
Snowball sample
Probability Samples

A means that allows us to know in advance the
likelihood that an element will be selected from
the population
 Relies on random selection
 Problems to watch out for when selecting a
random sample:
 An incomplete sampling frame
 Failure to obtain an adequate response rate
 Random samples and sampling error
 Generally, a random sample has sampling
error due to chance
 Use inferential statistics to calculate
sampling error
 2 things effect the degree of error due to
chance:
 The size of the sample
 The Homogeneity of the population
Probability Sampling Continued


Importance: Alf Landon vs. Roosevelt
presidential sample
Types of probability samples




Simple random sample (SRS)
Systematic random sample
Stratified random sample
Cluster sample
Sampling Distributions

Errors may occur when drawing samples (the
sample is not representative of the
population)

Two reasons why this can occur



You make a mistake (systematic sampling error)
Errors due to chance (random sampling error)
Use inferential statistics to determine sampling
error

Confidence Intervals

Usually reported at 95%, 99%, & 99.9%
Determining Sample Size



Usually use about 1,000 - 1,500 for U.S.
population if looking for a simple
description
Usually use up to 2,500 if wanting to
know about something detailed
Locally or regionally, typically a few
hundred