Measuring the dynamics of personality

Personality within individuals:

estimating individual differences in

endogenous and exogenous

affective responses

or

Measuring the dynamics of personality

William Revelle1, Eshkol Rafaeli2, Jon Adler1

1Northwestern University 2Barnard College, Columbia University

International Society for the Study of Individual Differences,

Graz, Austria, July, 2003http://www.personality-project.org http://pmc.psych.nwu.edu


• We can recognize a person by his or her pattern of affect,behavior and cognition just as we recognize a musicaltune that has a coherent pattern over time, even thoughplayed with different instruments in a different key, in adifferent tempo.

• Personality is the coherent temporal pattern of Affect,Behavior, Cognition and Desire.

• “Personality is a self-tunable system of the temporalpatterning of the interplay between affect, motivation,cognition, and behavior.” (Ortony, Norman, & Revelle, in press,Effective Functioning: A Three Level Model of Affect, Motivation, Cognition,and Behavior. To appear in J. M. Fellous & M. A. Arbib (Eds.), Who Needs

Emotions? The Brain Meets the Machine. New York: Oxford University Press.)


• Personality is the coherent temporalpattern of Affect, Behavior, Cognitionand Desire

• Focus today will be on dynamics ofAffect within subjects and the feedbackeffect of Behavior upon current andsubsequent affect.


• Structure of affect across individuals

• Structure of affect within individuals

• Sources of within subject variation



– Positive Affect (PA) and Energetic Arousal (EA)

– Negative Affect (NA) and Tense Arousal (TA)


– Large variation in correlation of PA and NA

– Not due to artifact of sampling


– Endogenous rhythm of EA/PA

– Exogenous “shocks” and NA/TA

Multiple formulations of the

measurement of affect• Two dimensional models

– Affective Valence and Arousal (Russell et al.)

– Positive and Negative Affect (Tellegen, Watson & Clark)

– Energetic and Tense Arousal (Thayer)

• Multidimensional models

– Pleasantness-unpleasantness, rest-activation, relaxation-attention(Wundt)

– Energetic Arousal, Tense Arousal, and Hedonic Tone (Matthews)

– Hierarchical models (Watson and Tellegen)

Between subject structure of affect

• Motivational State Questionnaire (MSQ)administered to ! 3500 subjects (datacollected over ! 10 years for many differentexperiments)

• Items from Thayer, Watson, Larsen & Diener

• 68-72 items administered at various times ofday and under various conditions.

• 2, 3, 4, 5, and 6 factor solutions

2 Dimensions of Affect

-1.0 -0.5 0.0 0.5 1.0-0.5

0.0

0.5

1.0

ENERGETI

ELATEDACTIVE

EXCITEDSTRONG VIGOROUSAROUSED

INSPIRED

DETERMIN

INTENSE

NERVOUS

SCAREDFEARFUL

AFRAID

TENSECLUTCHED

SORRY

DISTRESS

ANGRY

FRUSTRATUPSET

SADBLUE

UNHAPPYDEPRESSED

IRRITABL

GLOOMY

GROUCHY

CONTENT

SATISFIE

CONFIDENHAPPY

WARMHEARPLEASED

SOCIABLECHEERFUL

DELIGHTEWAKEFUL

INTEREST

ATTENTIVWIDEAWAK

ENTHUSIAPROUD

FULL_OF_

LIVELY

ALERT

ANXIOUS

AT_EASE

ASHAMEDHOSTILE

RELAXED

SLUGGISH

LONELY

GUILTY

TIRED

DULL

SLEEPY

JITTERY

DROWSY

INACTIVE

ASTONISH

AT_REST

CALM

SURPRISE

SERENETRANQUIL

BORED

QUIET

IDLE

STILL

PLACID

QUIESCEN

Energetic Arousal/Positive AffectEnergetic Arousal/Positive Affect

Ten

se A

rousa

l/N

egat

ive

Aff

ect

2 Dimensions of Affect Positive x Negative Affect

R = .04 N !3,200

Neg

ativ

e A

ffec

t

Positive Affect0 1 0 2 0 3 0

0

1 0

2 0

3 0











Studies of within subject affect

• Previously reported data from:

– Rogers (diary study over two weeks)

– Rafaeli (diary/PDA study over two weeks)

– Demonstrated stable individual difference insynchrony/asynchrony of affect

• Current focus will be on dynamic patterningacross time and over situations

Basic Method

• Subjects: Northwestern undergraduates

• 8-16 affect items given every 3 hours duringwaking day for 1 week, 1-2 week delay, and thenmeasured again for another week.

• Visual analog scales used by Rogers (N!80)

• Visual analog scales and Personal DigitalAssistants (Palm Pilots) used by Rafaeli (N!80)

• Will present data from the PDA portion ofRafaeli’s studies

Results: The correlation between

positive and negative affect

• Between subjects (all data points) r = -.04

• Between subjects (aggregated within subjects)

r = -.14

• Within subjects (all data points)

– -.5 < r < .7 (observed sd r = .27, expected= .14)

• Within subjects (aggregated by situations)

– -.8 < r < .9 (observed sd r =.43, expected = .33)

Positive and Negative Affect for all

subjects over all observations r= -.04Positive and Negative Affect (all observations)

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80

Positive Affect

Neg

ati

ve A

ffect

Mean Positive x Negative Affect

Aggregated within subjects r = -.14Postive and Negative Affect (subject averages)

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80

Positive Affect

Neg

ati

ve A

ffect

Large variation in within subject

correlation of positive and negative

affect (sd r =.27, expected =.14)

0 . 0 5

0 . 1 0

0 . 1 5

0 . 2 0

5

1 0

1 5

- 0 . 6 - 0 . 3 0 . 0 0 . 3 0 . 6 0 . 9

Within subject correlation

Fre

quen

cy

Positive and Negative Affect

(within subject correlation = -.51)F17 (r = -.51)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation = -.43)F03 (r = -.43)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation = -.09)E06 (r = -.09)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation =.11)G12 (r = .11)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation =.43)G08 (r = .43)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation =.55)D03 (r = .55)

0

20

40

60

80

0 20 40 60 80

PA

NA


(within subject correlation =.59)D05 (r = .59)

0

20

40

60

80

0 10 20 30 40 50 60 70 80

PA

NA

Large variation in within subject

correlation of positive and negative

affect (sd r =.27, expected =.14)

0 . 0 5

0 . 1 0

0 . 1 5

0 . 2 0

5

1 0

1 5

- 0 . 6 - 0 . 3 0 . 0 0 . 3 0 . 6 0 . 9

Within subject correlation

Fre

quen

cy

Measuring the dynamics of

personality

• Measurements taken every 3 hours during waking

day

• We can observe within subject change in positive

and negative affect over a day, replicated each day

• Note that while people differ in their pattern,

within each person, the pattern is consistent

Within subject dynamics (r = -.51) Within subject dynamics (r = -.43)

Within subject dynamics (r = -.05) Within subject dynamics (r = .55)

Within subject dynamics (r=.59)Measuring the dynamics of personality










Endogenous rhythms of affect

• Possible to fit each person’s daily data with

cosine of time of day to estimate any

diurnal pattern.

• Positive affect and energetic arousal show

strong diurnal rhythm

• Negative affect and tense arousal show very

weak diurnal rhythm

Fitting affect within subjects

Fitting affect within subjects Fitting affect within subjects

Fitting affect within subjects Fitting affect within subjects

Fitting affect within subjectsMeasuring the dynamics of personality










Affect varies by situation

• A small effect of situation on affect, but a

much larger individual difference in

experience of affect by situation

• Is the affect a function of the situation, or is

the situation a function of prior affect?

• Individuals differ in their choice of situation,

as well as the affect they experience in the

situation

Individual affect depends upon

person in the situation

• Situations were defined using Dan Ozer’s

categorization of typical student activities

• Aggregate the affect ratings by situations to

examine the effect of situation on

individuals.

• Note the variation between subject in the

effect of situations

A small effect of situations on

positive and negative affect

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00

sleep relax

exerciseromantic

friends

work

classstudy

r = .695D03 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

sleep relax

groom

motion

study

class

workexercisefriends

familyrecreateeatchores

D18 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

motion

recreate

exercise

class

friendseat

study

groomwork

sleeprelax

chores

r = .571E16 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

romantic

motion

exercise

work

class

groom

friends

eat

studyrelax

choressleep

r = .530

r = .505D11 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

sleepexercise

romantic

rather not say

groomfamily

errandfriends

motion

relaxeat

studyclassrecreate


0

20

40

60

80

0 20 40 60 80

PA

NA work

errand

exercisechores

motion friends

classgroom

eatstudy

relaxsleep

r = .441


0

20

40

60

80

0 20 40 60 80

PA

NA

motion

classerrandexercise

eat friendswork

romanticrecreation

groomsleep

studyrelax

other

r = .191 r = -.433E19 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

exercisework

groom

errandrelax

sleep

family

study eat

friendsclassother

r = -.539D01 Across Activities

0

20

40

60

80

0 20 40 60 80

PA

NA

motion

friends

errand

eatrelax

study

sleepwork

class

groom recreate












If personality reflects stability in the changingpattern of Affect, Behavior, and Cognition acrosstime, then we need to focus on measuring Affect,Behavior, and Cognition within subjects acrossshort and long temporal intervals.

With the ability to use new data collectiontechniques (e.g. PDAs) and new analytictechniques (e.g., HLM) we can measure thedynamics of personality within individuals as wellas between individuals.

Documents

Measuring the dynamics of personality