Transcript Section 5.2 – Designing Experiments

SECTION 5.2 –
DESIGNING
EXPERIMENTS
VOCAB

Experimental Units –


Subjects –


Human beings for experimental units.
Treatment –


Individuals on which the experiment is being done.
Specific condition applied to an experimental unit.
Factors –

Explanatory variables.
VOCAB

Placebo –


Placebo Effect –


Psychological effect of responding favorably to any
perceived treatment.
Control Group –


Dummy Pill/Treatment.
Group receiving no treatment or placebo treatment.
Experimental Group –

Group receiving experimental treatment.
THREE BASIC PRINCIPLES OF
EXPERIMENTAL DESIGN
1.
Control –

2.
Randomize –

3.
Control the effects of lurking variables on the
response, most simply by comparing 2 or more
treatments.
Use chance to assign units to treatments.
Replicate –

Use the same treatment on many units to reduce
chance variation.
COMPLETELY RANDOMIZED DESIGN

A completely randomized design happens when
all experimental units are allocated at random
among all treatments.
EXAMPLE #1:

1.
What is the best way to answer each of the
questions below: an experiment, a sample survey,
or an observational study? Explain your choices.
Are people generally satisfied with how things
are going in the country right now?
o
Survey, no treatment, and cannot tell if someone is
satisfied by observing them.
EXAMPLE #1:
2.
Do college students learn basic accounting
better in the classroom or using an online
course?
o
3.
Experiment, has two treatment that results can be
compared.
How long do your teachers wait on the average
after they ask their class a question?
o
Observational Study, no treatment given, and
responses to survey questions would be bias.
EXAMPLE #2:

New varieties of corn with altered amino acid
content may have higher nutritional value than
standard corn, which is low in the amino acid
lysine. An experiment compares two new
verities, called opaque-2 and floury-2, with
normal corn. The researchers mix corn-soybean
meal diets using each type of corn at each of
three protein levels, 12% protein, 16% protein,
and 20% protein. They feed each diet to 10 oneday-old male chicks and record their weight gains
after 21 days. The weight gain of the chicks is a
measure of the nutritional value of their diet.
EXAMPLE #2:

What are the experimental units and the
response variable in this experiment?

Experimental Units – one-day-old male chicks

Response Variable – weight gain
EXAMPLE #2:

How many factors are there? How many
treatments? How many experimental units does
the experiment require?

Factors – 2, amino acid content and protein level.

Treatments – 6, 2 different types of amino acid with 3
levels of protein for each type.

Number of experimental units – 60, 6 treatments
with 10 chicks each.
EXAMPLE #2:

Use a diagram to describe a completely
randomized design for this experiment.
60
Chicks
Group 1
10 Chicks
Treatment 1
Opaque-2, 12% P
Group 2
10 Chicks
Treatment 2
Opaque-2, 16% P
Group 3
10 Chicks
Treatment 3
Opaque-2, 20% P
Group 4
10 Chicks
Treatment 4
Floury-2, 12% P
Group 5
10 Chicks
Treatment 5
Floury-2, 16% P
Group 6
10 Chicks
Treatment 6
Floury-2, 20% P
Compare
Weight
Gains
MORE VOCAB

Double Blind–


Matched-Pairs Design –


Compare just two treatments. A design to compare
two treatments through one-sample procedures.
Block –


Neither subject nor the people that have contact with
them know the treatment being used.
A group of experimental units that are known to be
similar in some way that is expected to affect the
response to the treatments.
Block Design–

In a block design, the random assignment of units to
treatments is carried out separately within each
block.
EXAMPLE #1:

Do consumers prefer the taste of a cheeseburger
from McDonald’s or from Wendy’s in a blind test
in which neither burger is identified? Describe
briefly an experiment that could be used to
answer this question.
o
Make the experiment a combination of double-blind
and matched pairs procedures. The source of the
cheeseburger can’t be know ahead of time by the
tasters, it could bias their opinion. Each subject
must taste both cheeseburgers in order to identify
with one they prefer.
EXAMPLE #2:

Is the number of days a letter takes to reach
another city affected by the time of day it is
mailed and whether or not the zip code is used?
Briefly describe the design of a two-factor
experiment to investigate this question. Be sure
to specify the treatments exactly and to tell how
you will handle lurking variables such as day of
the week on which the letter is mailed.
EXAMPLE #2:

Use a diagram to describe this experiment.
X Pieces
of Mail
Group 1
X/n Mail
Treatment 1
Zip given, Time 1
Group 2
X/n Mail
Treatment 2
Zip given, Time 2
Group 3
X/n Mail
Treatment 3
Zip given, Time 3
Group 4
X/n Mail
Treatment 4
No Zip, Time 1
Group n-1
X/n Mail
Treatment n-1
No Zip, Time 2
Group n
X/n Mail
Treatment n
No Zip, Time 3
Compare
number of
days
EXAMPLE #2:

Things to think about.

Since the only factors that we are concerned about
are:
Time of day, and
 Zip Code (Yes or No), then

All letters need to be the same size
 All letters need to be sent to the same city.
 Mail letters on same day of week, but not around
holidays.
