Transcript The Language of Sampling

The Language of
Sampling
Lecture 5
Sections 2.1 – 2.4
Fri, Sep 8, 2006
Why Sample?



Studying a sample gives us only partial
information about a population.
So why not study (observe) the entire
population?
Samples are random, so how can we expect a
sample to be representative of the population?
Why Sample?

We can prove mathematically that a large
random sample has a very high probability of
being representative of the population from
which it is chosen.
The Language of Sampling






Unit or subject.
Variable.
Population size N.
Sample size n.
Parameter.
Statistic.
Parameters and Statistics

For numerical data, we usually use the average
of the values in the sample.

For example, the average household income.


What is the variable?
For non-numerical data, we usually use the
proportion of observations in a specific
category.

For example, the unemployment rate.

What is the variable?
Random vs. Representative


Random sample.
Representative sample.
Example

Study: Men Enjoy Watching Bad Guys Suffer
What were the populations?
 What were the samples?
 What were the variables?
 What statistics were used?
 What were the parameters?

Bias


A sampling method is biased if it systematically
produces a sample whose characteristics differ
from those of the population.
Note that it is the method that is biased, not the
sample.
Two Biased Sampling Methods



Convenience sampling.
Volunteer sampling.
Examples?
Three Types of Bias





Selection bias.
Nonresponse bias.
Response bias.
Experimenter bias.
Examples?
Whose Fault is it?




Selection bias originates in the sampling
procedure.
Nonresponse bias originates in the subjects who
were selected for the sample, but chose not to
participate.
Response bias originates in the subjects who are
in the sample.
Experimenter bias originates in the
experimenter.
Examples


Phone surveys.
Use random-digit dialing.
Convenience sampling?
 Volunteering sampling?
 Selection bias?
 Non-response bias?
 Response bias?
 Experimenter bias?

Examples


Mailed surveys, including e-mail.
Mail individuals a survey and ask them to
respond.
Convenience sampling?
 Volunteering sampling?
 Selection bias?
 Non-response bias?
 Response bias?
 Experimenter bias?

Examples


Internet survey.
Post the survey questions on the internet and let
visitors respond at will.
Convenience sampling?
 Volunteering sampling?
 Selection bias?
 Non-response bias?
 Response bias?
 Experimenter bias?

Examples


Estimating average family size.
Randomly select individuals and ask them how
many siblings they have.
Convenience sampling?
 Volunteering sampling?
 Selection bias?
 Non-response bias?
 Response bias?
 Experimenter bias?

Tomorrow
Bring
your calculator!