Moderators and Mediators

1
Moderators and Mediators
Class 7
2
Breaking News Mud Improves Memory     FLASH!
Explorers in the Amazon have discovered an
amazing memory mud. When smeared on the
forehead it vastly improves short term memory.
However, it appears that mem-mud works only for
certain kinds of people in certain places at
certain times. The mud appears to have its
powerful effects because it SIGNAL LOST.      
What questions arise?
Who does it work for, under what conditions, at
what times?
Why does it work?
3
Moderators, Mediators, and Mem-mud

• Moderator What are the conditions under which
  Mem-mud works, and for whom
• Mediator Why does Mem-mud work
• Moderator The limiting conditions that
  determine and effect or outcome (e.g., when, or
  for whom)
• Mediator The underlying cause Why something
  occurs.

4
MODERATOR DEFINED   A moderator is a variable
that affects the direction or strength of the
relationship between a predictor (IV) and an
outcome (DV). A ? B, if C is a moderator
model.   Changes in the moderator will affect
the relation between IV and DV.   Moderators
directly explain how and when events occur (but
can indirectly indicate why they
occur).   Moderator can be qualitative
variable (race, sex, etc.) quantitative
variable (perceived risk, mood)
5
MODERATOR DEFINED, Cont.
Moderator is always an Independent
Variable   Moderator should be un-correlated to
DV   Moderators sought when there is weak or
inconsistent relationship between IV and
DV.   Moderator confirmed only if interaction
between moderator and main IV is significant.
6
Moderation Correlation Between Variables is
High in One Condition But Not in Another.
These two results say what about bad events and
illness?
Major life events ? increased illness, IF event
is uncontrollable.
Moderator is?
CONTROLLABILITY
7
MODERATOR MODEL
Predictor (IV)
Moderators (IV)
Outcome Variable (DV)
Predictor X Moderator
Must be significant
Recipient race (IV)
Self-image risk(IV)
Outcome Variable (DV)
Race X Risk
p lt .005
8
Feedback Bias With and Without
Content/Mechanics as Moderator
9
Moderator Recipes

• a. IV is qualitative (dichotomous, polychotomous,
  Mod. is qualitatitve (dichotomous, polychotomous)
• b. IV is qualitative, Mod. is quantitative
• c. IV is quantitative, Mod. is qualitative
• d. IV is quantitative, Mod. is quantitative

10
Social Context, Disclosure, and Baby Cries
Primary IV?
Social Contact
Moderator
Disclosure
DV?
Distress Rating
IV is Qualitative, and Moderator is Qualitative
Recipe?
11
Height Perception as Function of Nature of Social
Contact and Duration of Social ContactAdapted
from Schnall, Harber, Stefanucci, Proffitt,
2008 Results amplified for purposes of
illustration
Primary IV?
Social Contact
Moderator
Duration
DV?
Hill slant
IV is Qualitative, and Moderator is Quantitative
Recipe?
12
Primary IV?
Depression
Moderator
Gender
DV?
Endorsement of Directive Support
IV is Quantitative, and Moderator is Qualitative
Recipe?
13
Self Esteem and the Use of Affect as
Information (Harber, 2005)   Mild upset 1 SD lt
mean, Moderate upset mean, Extreme upset 1
SD gt mean.
Primary IV?
Upset
Moderator
Esteem
DV?
Baby Cry Ratings
IV is Qualitative and Moderator is Qualitative
Recipe?
14
MEDIATION   1. History S ? R explanations do
not address full range of psychological
phenomena. In many cases, S?O?R makes more
sense.   2. Function Mediational analyses
designed to test for S?O?R causal paths.   3.
Mediation Defined the effects of stimuli on
behavior are mediated by various transformations
internal to the organism.   4. Utility
Mediators directly explain why events occur (but
can also provide clues to how and when they
occur).   5. Character Mediators are latent
variables, or latent constructs.
15
Attributes of Mediators   1. Changes in IV
account for changes in mediator   2. Changes in
mediator account for changes in DV.   3. When
links between IV to Mediator, and between
Mediator to DV are established, the link between
IV and DV becomes non-significant, or becomes
significantly decreased.  
16
Mediational Model
Mediator
a
b
DV
c
IV
When Mediation is present r (a) is
significant r (b) is significant r ( c ), which
was significant before the mediator was
included, is either not significant or is
much weaker after mediator is
included.
17
Mediational Model and Feedback Studies
NOTE Expected result after getting ind. diff.
measure of egalitarianism.
18
RELATION BETWEEN MODERATORS AND
MEDIATORS     Moderator to mediator Knowing how
external variables affect outcomes can imply the
existence of mediators.   Mediator to Moderator
Knowing why underlying mechanism affects outcomes
can suggest ways of addressing, remedying,
altering the outcome.
19
Why Do People Need Self Esteem? Converging
Evidence that Self Esteem Provides an
Anxiety-Buffering Function   Greenberg, et al.,
2000    Three studies were conducted to assess
the proposition that self-esteem serves an
anxiety-buffering function. In Study 1, it was
hypothesized that raising self-esteem would
reduce anxiety in response to vivid images of
death. In support of this hypothesis, subjects
who received positive personality feedback
reported less anxiety in response to a video
about death than did neutral feedback subjects.
. .
X
____Moderator
____Mediator
20
Greenberg, et al. 2000
Scary Images

Self Esteem
0
Anxiety

Scary Images X Self Esteem
21
Anxiety as a Function of Threat Salience and
Level of Self Esteem   Greenberg, et al., 2000
22
Visceral Perception and Nonconscious Fear
Conditioning   Katkin, Wiens, Öhman,
2001   Previous research shows that people
conditioned to fear certain kinds of stimuli such
as snakes or spiders are subsequently better able
to detect hidden images of these stimuli compared
to people who are not conditioned. The current
research predicts that this heightened
sensitivity is restricted to people who are good
at detecting their own heartbeats (good heartbeat
monitors). This prediction was confirmed when
good heartbeat monitors are excluded from
analysis the effect of fear conditioning on
stimuli sensitivity disappears. ABSTRACT
MODIFIED
X
____Moderator
____Mediator
23
Katkin, Wiens, Öhman, 2001
Heartbeat Monitors
a
b
c
Stimuli Sensitivity
Conditioned Fear
24
Path Analyses and Causal Models

• Attend. in 21st Century program ? improved
  performance by minority students.
• 21st Century program based on theory of
  stereotype-threat. When threat is high, minority
  students do less well.
• Stereotype threat is itself a problem because it
  causes minority students to disidentify with
  academics.
• What is the causal model? What would be strong
  and weak correlational links in this model?

25
First Semester Grades as Predicted by Stereotype
Threat, Identification with School, and
Participation in 21st Century Program
ID with School
Stereo. Threat
21st Cent. Prog.
Grades
Task What are IVs, DVs? What are
moderators (if any)? What mediators (if any)?
Put these in correct path
Estimate relations between variables, outcome
26
21st Century Program, Stereotype Threat, and
Identification with School
27
Ultimate Mediation? The Epistemological
Challenge of Science    Multiple underlying
causes   Disclosure --gt Emotional resolution --gt
less stress --gt immune boost --gt health   Race of
recipient --gt egalitarian concerns --gt self-image
concerns --gt bias     Ultimate underlying
causes   Disclosure ---gt coping (Pennebaker,
1989) Disclosure --gt ???? --gt coping Disclosure
--gt self-affirmation --gt coping (Cresswell, et
al., 2007) Disclosure --gt self-affirmation --gt
???? --gt coping  
28
(No Transcript)
29
(No Transcript)