This preview shows pages 1–3. Sign up to view the full content.
Section V.
Studying Your Interests:
Data Gathering
1. Selecting Units to Observe: Subjects and Sampling
Survey research is very concerned about representativeness (external validity), so
sampling
procedures are very important. Experimental research is more concerned about demonstrating
causal relationships so control of extraneous variables (external validity) is of primary
importance; usually
subjects
are obtained using nonprobability techniques.
A. Probabilistic Sampling
Involves random sampling techniques
Allows generalizations from a sample to a population with a known amount of error at a
certain level of certainty (we are 95% certain that the sample estimate is off by no more
than ± 3 percentage points)
The most effective means for selecting representative subjects
Avoids researcher biases. Population units are selected by chance and not by the
researcher.
Permits for estimates of sampling error
Requires a sampling frame  a listing of everyone (every object) in the population.
Sample units are selected from this list such that all members of the list (sampling frame)
have an equal and independent chance of being selected, i.e., units are selected randomly
 chance determines which population units end up in the sample
Nonresponse bias (nonsampling error) is always a problem. If only a portion of the
random sample responds, is the nonresponse random, or do the nonrespondents
constitute a particular segment of the intended population? Nonresponse can bias the
results.
Always compare sample characteristics (demographics, etc.) to known population
characteristics to check for nonresponse bias
Probabilistic sampling techniques:
Simple random sampling
Assign a single number to each element, not skipping any number in the process
Use random number table to select elements for the sample
Systematic sampling
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Every nth element in the total population list is selected for inclusion into the sample with
a random starting point
Stratified sampling
Grouping members of a population into homogeneous strata before sampling
Reduces degree of sampling error
Multistage cluster sampling
Used when population list does not exist
Sample of members (clusters) is selected
Members of the selected cluster are listed
List of clusters is subsampled
B. Nonprobabilistic Sampling
Does not rely on random sampling techniques
Should not be used to make generalizations to a larger population; while a purposive
sample may be generalizable, you cannot be as sure as with random sampling
When it is either impossible or unfeasible to select a probabilistic sample i.e., when there
is no sampling frame of the population (e.g., drug abusers)
Less valid than probabilistic sampling
Practical
Nonprobabilistic sampling techniques:
Purposive Sampling: researcher uses own judgment in selection of members
Chunk Sampling: a convenient group of people (a class of students etc.) are used as a
This is the end of the preview. Sign up
to
access the rest of the document.
 Fall '06
 EdwardWotring

Click to edit the document details