Lecture 11: Experimental Material Three Decisions

advertisement
Lecture 11: Experimental Material
Three Decisions
Three Decisions
Chapter 4 looks at the Three
Decision discussed earlier
in more detail.
Response?
Conditions?
Experimental Material?
1
2
Experimental Material
Experimental Material
Experimental material will
be the objects or items that
will be experimented on.
An experimenter measures
the response on the
experimental material.
3
Experimental Material
4
Experimental Units
Often times the
experimental material is the
same as the experimental
units.
Treatments are assigned to
experimental units.
5
6
1
Lecture 11: Experimental Material
Example
Example
Conditions: Regular hog feed,
Regular hog feed with Paylean
(an additive advertised to add
lean weight more quickly to
the hog).
Response: Weight gain (lbs)
Experimental Material:
young hogs.
7
Example
8
Example
Hogs are raised in pens. There
are around 20 hogs to a pen.
Each pen has a common feed
trough where the hogs compete
for the food. Each pen gets the
same amount of feed per day.
The treatment (type of feed:
Regular or with Paylean) is
assigned at random to pens.
9
Example
10
Example
Because hogs compete for a
fixed amount of food for the pen,
if some hogs get more other hogs
will get less. The weight gains
of the hogs within a pen are not
independent.
Experimental material:
individual hogs.
Experimental units: pens
with 20 hogs.
11
12
2
Lecture 11: Experimental Material
Example
Alternative Example
Because the pens are the
experimental units. The
actual response should be
the average weight gain for
the pen.
Hogs are fed individually.
The same amount of food is
given to each hog each day.
13
Alternative Example
14
Alternative Example
Because treatments
(Regular or Paylean) will be
assigned to individual hogs,
individual hogs are the
experimental units.
Experimental material:
individual hogs.
Experimental units:
individual hogs.
15
Alternative Example
16
General Goals
Have experimental material
that is representative.
If experimental material is
representative, the results of
the experiment can be
generalized.
Hogs are treated
individually and so their
weight gains are
independent observations.
17
18
3
Lecture 11: Experimental Material
General Goals
General Goals
These two goals are often
competing.
Have experimental material that
is uniform.
If experimental material is
uniform there is not much
natural variability, making
treatment effects easier to detect.
More representativeness often
means less uniformity.
More uniformity often means
less representativeness.
19
Representative
20
Random Sampling
Random sampling from a larger
population of experimental
material does not guarantee
representative experimental
material but it does decrease the
chance that the experimental
material will be biased.
A representative set of
experimental material
would look like the larger
population from which it
was taken.
21
22
Control
Experimental Material
Control of outside variables can
make experimental material
more uniform.
By not letting an outside variable
change, the experimental
material will be uniform with
respect to that variable.
Random selection – less
uniform – less chance of being
biased.
Control of outside variables –
more uniform – less
representative.
23
24
4
Download