Types of Relationships - Robert H. Gass

Download Report

Transcript Types of Relationships - Robert H. Gass

Correlations between variables
Differences between groups and
conditions
Types of Relationships
• Social scientists are interested in
discovering functional relationships
between variables.
• In particular, researchers look for:
– correlations (association,
covariation) among vaariables
– differences between groups or
conditions
The nature of causation
• Cause-effect relationships--causation is
always inferred, never directly observed
• “functional” relationships
– one thing correlates with, or is associated with
another (correlation)
– one thing predicts or explains the amount of
variance in another (analysis of variance)
– one thing has a direct effect on another (path
analysis, multiple regression)
Graphic Representations of
Relationships
(dependent
variable)
Y-axis
(independent variable)
X-axis
Note: “3/4 rule”
the convention is
to make the Y axis
3/4 of the length
of the x axis
Correlations
• displaying correlations using a scattergram
• linear relationship
– can be positive or negative
• curvilinear relationship
– also known as nonmonotonic relationships,
quadratic trends, “u-shaped” or “inverted-u”
– requires a minimum of three levels of the
variable being investigated
• no correlation
• spurious effect
Do employees who drink a lot of coffee take more
bathroom breaks?
frequency table
employee cups of
coffee
bathroom
trips
Fifi
1
2
Rex
2
1
Trudy
3
3
Pedro
3
4
Lulu
4
6
Thaddeus
5
5
Rudolfo
6
5
scatterplot
Illustration of Scatterplots
• Scatterplots that are
closer to a straight
line have correlations
closer to +1.0 or -1.0
• Must have interval or
ratio data
• Correlation does not
prove causation
Linear versus curvilinear relationships
Linear relationship
Curvilinear relationship
Differences Between Groups
or Conditions
• main effect (changes produced by one
independent variable alone)
– one-way interaction
• interaction effect (changes produces by
independent variables acting together, or
in concert
– two-way interaction
– three-way interaction
interpersonal touch, social labeling, and
the foot-in-the-door effect
touch
no touch
positive
FITD
.15
.40
negative
FITD
.45
.25
communicator physical attractiveness
and persuasion
attractive
criminal
swindler
burglar
unattractive
criminal
5.45
4.35
2.80
5.20
non-significant-interaction
A characteristic feature of
non-significant interaction
effects is that the lines are
parallel, or nearly parallel
potentially significant interaction
(ordinal)
potentially significant interaction
(disordinal or “crossed”)
potentially significant interaction
(ordinal)
Illustration of an interaction effect
2
3
3
6
Attitude change
0 1 2 3 4 5 6 7
Source Credibility
low
high
evidence quality
low
high
high quality
evidence
low quality
evidence
low
high
Source Credibility
HOT
ROOM
900
HARD
TEST
60%
60%
EASY
TEST
60%
80%
0 10 20 30 40 50 60 70 80 90 100
COLD
ROOM
500
Test score
Illustration of an interaction effect
Easy test
Hard test
500
900
What type of effect is this?
12
10
8
Low
Esteem
6
4
mean
speech
errors =
9.5
2
Low
mean
Esteem speech
Speaker errors =
6.1
mean
speech
errors =
2.2
High
Esteem
0
High
mean
Esteem speech
Speaker errors =
2.1
Audience
Present
Average number of speech errors
No
Audience
Present
no
audience
audience
present
What kind of correlation?
• Are people in richer nations happier??