Factorial designs
Download
Report
Transcript Factorial designs
Single-Factor (IV) designs
•
All these designs we’ve talked about so far have
investigated the effects of ______ IV (_______)
or multiple levels of the ______ IV
•
But, research questions and hypothesis often
involve multiple IV’s, and various levels of each
IV
1
Factorial Designs
________________________ investigate the effects of
multiple independent variables
These studies investigate Multiple _______ (IVs) and
________ (variations) of those IVs
e.g., Participants in a two-factor design are exposed to
particular level of factor A __________________ with
particular level of factor B Factor B
Level B1
Level B2
Level A1
Condition
A1B1
Condition
A1B2
Level A2
Condition
A2B1
Condition
A2B2
Factor A
2
Factorial Designs
You can use more than 2 factors, and more than two
levels, but your study can get _____________,
Expressed in terms of Factor 1 levels __Factor 2 levels,
etc.
How many numbers = how many _________
Value of number = how many ________
e.g., a 2 x 2 design has __ factors and __ levels of
each
e.g., a 2 x 2 X 4 design has __ factors, with the first
__ factors having __ levels of the IV, the third
having ___ levels of the IV
3
Factorial Designs
Allows you to investigate how multiple variables ______
Functional relationship is demonstrated when you see
the _____________________________.
Require many groups (2 x 2 = __ groups, 2 X 3 = __
groups)
Can detect 2 kinds of effects
_____________: a factor (IV), regardless of level of
other factor (IV), produces an effect
______________: a factor (IV) only produces an
effect at one level of the other factor (IV)
4
Factorial Designs: Example
Variables influencing risk taking:
Factor (IV):
A. Context:
Gain or Loss
B. Outcome type:
Hypothetical or Real
Level (variations of IV)
A1. Money gains
A2. Money losses
B1. Hypothetical money
B2. Real money
Factor A = Context
Factor B =
Outcome
Type
Gain
Loss
Hypothetical
Gain/Hyp
Loss/Hyp
Real
Gain/Real
Loss/Real
5
____ main effect, ____ interaction
A
Risky Choices
100
80
Hypothetical
Real
60
40
20
0
Gains
Losses
Context
Gain
Loss
Hyp.
Gain/Hyp
5
Loss/Hyp
6
Real
Gain/Real Loss/Real
6
5.5
B
11
11
11.5
11.5
6
Main effect, no interaction
________________:
Gain
Loss
Hyp.
Gain/Hyp
10
Loss/Hyp
85
48
Real
Gain/Real
9
Loss/Real
87
48
9.5
86
Risky Choices
100
80
Hypothetical
60
40
Real
20
0
Gain
Losses
Context
A
B
Main effect Factor ____
7
Main effect, no interaction
__________________:
A
B
Real
Gain
Loss
Gain/Hyp
85
Loss/Hyp
83
84
Gain/Real
15
Loss/Real
18
53
50
50.5
Risky Choices
Hyp.
100
80
Hypothetical
60
40
Real
20
0
Gain
Losses
Context
Main effect Factor ___
8
Main effect, no interaction
___________________:
A
Risky Choices
100
Gain
Loss
Hyp.
Gain/Hyp
65
Loss/Hyp
85
75
Real
Gain/Real
37
Loss/Real
58
48
51
71.5
80
Hypothetical
60
40
Real
20
B
0
Gain
Losses
Context
Main effect Factors ___________
9
No main effect, interaction effect
Risky Choices
80
A
60
Hypothetical
40
Gain
Loss
Hyp.
Gain/Hyp
25
Loss/Hyp
79
52
Real
Gain/Real
80
Loss/Real
25
53
52.5
52
Real
20
0
Gain
Losses
Context
B
Lines are not _______: Effects of one IV ________ on the
the level of another IV
10
Main effect, interaction effect, but
_____________ more important
A
Risky Choices
80
60
Hypothetical
40
Gain
Loss
Hyp.
Gain/Hyp
78
Loss/Hyp
77
78
Real
Gain/Real
77
Loss/Real
15
46
77.5
46
Real
20
B
0
Gain
Losses
Context
Two main effects but, lines are not ___________:
Effects of one IV depends on the level of another IV
11
Factorial Design w/ Repeated measures
A factorial design in which you measure performance in
subjects _____________
Controls for _____________________
Example: Experimenter is interested in investigating the
effects of a drug on performance rats given different
diets. Each rat is given both diets and exposed to each
drug dose.
12
Factorial Design w/ Repeated measures
Diet
Drug dose
Sugar rich
Protein Rich
0 mg
Sugar rich
0
Protein Rich
0
2 mg
Sugar rich
2
Protein Rich
2
4 mg
Sugar rich
4
Protein Rich
4
13
Mixed design- between/within subject
A factorial design in which you measure performance in
subjects _________________
Have at least one __________-subjects factor and one
___________-subjects factor
Example: Experimenter is interested in investigating the
effects of a drug on performance rats given different
diets. Each rat is given one diet type (between-subjects
variable) and exposed to each drug dose (within-subject
variable).
14
Mixed design- between/within subject
Diet
Group 1
Sugar rich
Group 2
Protein Rich
0 mg
Sugar rich
0
Protein Rich
0
2 mg
Sugar rich
2
Protein Rich
2
4 mg
Sugar rich
4
Protein Rich
4
Drug dose
Drug is the ______subject variable, Diet is the ______ subject variable
15
Subject Variable
In some research designs, quasi-experimental
research, investigators try to relate _________
variables or non-manipulatable variables to
behavior
Subject variable = __________ characteristic of
individuals, e.g., gender, personality, presence
of a disorder, age, IQ,
___________ because it is not manipulated
You cannot attribute results to subject variable
because other variables were ____________
16
Factorial Design w/ Quasi-Independent
(Subject) variable (PXE)
Quasi-independent or Subject variable – Variable
that ____________ in the population and is not
______________, or ______.
Example:
Factor A. Outcome type
Classification variable (non-manipulated): gender
Gender
Outcome
Type
Male
Female
Hypothetical
Male/Hyp
Female/Hyp
Real
Male/Real
Female/Real
17
TEST your knowledge
1. A factor is another name for what?
2. A 2 X 3 X 2 factorial design has how many
IV’s? How many levels of each IV
3. Draw the results of a factorial design study
showing 1 main effect, an interaction.
Describe the design of this study:
In a study of circadian variability in skin sensitivity, a
researcher examines 2-point thresholds for 20
participants. Each is tested on three different areas of
the body (thumb, forehead, thigh) at three different
times of the day (9 AM, 3 PM, 9 PM, and 3 AM). The 2point threshold is the distance between two points (in
cm) touching the skin, at which a participant perceives
them as two points instead of one.
______________________
In a study of crowding, introverts and extraverts are
to be compared. Some are tested in a large room (8’
x 8’), others are tested in a medium-sized room (6’ x
6’) and others are tested in a small room (4’ x 4’). To
see if expectations will influence the outcome, half
the participants in each group are told that the study
hypothesizes that crowding will harm their
performance; the remaining participants are told
nothing about hypothesis being studied. While in the
room, participants fill out an anxiety survey
(psychological test), then listen to a series of brief
segments of five heavy metal songs and rank them
from the most annoying to the least annoying
_________________________________________