The Couples

Satisfaction Index (CSI)

Ronald D. Rogge
Asst. Professor of Psychology
University of Rochester

[email protected]
www.couples-research.com
Overview
 PART 1: Development of CSI
 Existing scales
 Development of new scale
 Cross-sectional validation
 Longitudinal validation

 PART 2: Use & Interpretation of CSI

 Administration
 Scoring
 Interpretation
 Norms
PART 1: Existing Scales
 Strengths
 20-30 years of converging results
 Clearly measure satisfaction

 Limitations
 20-30 years old
 Heterogeneous content

 Unknown noise
Existing Scales

Scale Items Name Cit. Cit./Yr

DAS 32 Dyadic Adjustment Scale 2,191 77.1

MAT 15* Marital Adjustment Test 1,489 32.1

QMI 6 Quality of Marriage Index 221 9.9

Relationship Assessment
RAS 7 Scale 156 8.8
Evaluating Scales
 Item Response Theory
 Used to create SAT, GRE, MCAT

 Item by item analysis

• If happy, higher responses?
• If unhappy, lower responses?

 Requires large samples

• Estimates parameters for each item
• Estimates parameter for each subject

 Sample-Independent Results
 DAS-31
(Please indicate the degree of happiness, all things considered, of your relationship.)

1 - Extremely Unhappy
2 - Fairly Unhappy
1 4.5
3 - A little unhappy
4 - Happy
0.9 4
.

5 - Very Happy
6 - Extremely Happy

.
0.8
7 - Perfect 3.5
Probability of each response

0.7
3

Information (theta)
0.6
2.5
0.5
2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Satisfaction Satisfaction
 DAS/MAT 5
Agreement on: FRIENDS

1 - Always Disagree
.

1 4.5
2 - Almost Always Disagree

0.9 3 - Frequently Disagree 4

Probability of each response

.
4 - Occasionally Disagree
0.8
3.5
5 - Almost Always Agree

0.7 6 - Always Agree

3

Information (theta)
0.6
2.5
0.5

2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Satisfaction Satisfaction
Study 1: Goals
 Evaluate current scales

 DAS, MAT, QMI, RAS

 IRT in large sample

 Develop CSI

 Large item pool

 Factor analysis
 IRT
Study 1: Method
 Online survey (N = 5,315)

 Contents
 141 satisfaction items
 Items from DAS, MAT, QMI, RAS
 71 additional items

 7 anchor scales
 e.g., neuroticism, hostile conflict, stress

 2 validity scales
Study 1: Sample
 Avg 26yo (SD=10yr)
 26% High School or less
 83% Female
 76% Caucasian
 Relationships
 24% Married (avg 6.3yrs)
 16% Engaged

 60% Committed dating

Relationship Quality

Sample Size Length of Satisfaction

(N) relationship (DAS)

Married 1254 9.0 yrs 108

Engaged 866 3.1 yrs 117

Dating 3194 1.7 yrs 113

Evaluating Previous Scales
 IRT results

 Evaluated 66 items of existing scales

 Some very informative items

 Many poor items

 DAS-31
(Please indicate the degree of happiness, all things considered, of your relationship.)

1 - Extremely Unhappy
2 - Fairly Unhappy
1 4.5
3 - A little unhappy
4 - Happy
0.9 4
.

5 - Very Happy
6 - Extremely Happy

.
0.8
7 - Perfect 3.5
Probability of each response

0.7
3

Information (theta)
0.6
2.5
0.5

2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Satisfaction Satisfaction

Standard Deviation (SD) Units Standard Deviation (SD) Units

 QMI-1
We have a good relationship

1 4

0.9
3.5
.

0.8
3

.
Probability of each response

0.7
1 Very Strong Disagreement
2.5
0.6 2

Information (theta)
3
0.5 2
4

0.4 5
1.5
6
0.3
7 Very Strong Agreement
1
0.2

0.5
0.1

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Satisfaction
Satisfaction
 SMD-2
BAD 1 – 2 – 3 – 4 – 5 – 6 GOOD

1 4.5
.

0.9
4

.
Probability of each response

0.8
3.5

0.7
1 - BAD 3

Information (theta)
0.6 2
3 2.5
0.5
4
2
5
0.4
6 - GOOD
1.5
0.3

1
0.2

0.1 0.5

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Satisfaction Satisfaction
 DAS/MAT 5
Agreement on: FRIENDS

1 - Always Disagree
.

1 4.5
2 - Almost Always Disagree

0.9 3 - Frequently Disagree 4

Probability of each response

.
4 - Occasionally Disagree
0.8
3.5
5 - Almost Always Agree

0.7 6 - Always Agree

3

Information (theta)
0.6
2.5
0.5

2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Satisfaction Satisfaction
 DAS/MAT 6
Agreement on: SEX RELATIONS

1 1 - Always Disagree 4.5

.

2 - Almost Always Disagree

0.9 4
3 - Frequently Disagree
Probability of each response

4 - Occasionally Disagree

.
0.8 3.5
5 - Almost Always Agree
0.7
6 - Always Agree 3

Information (theta)
0.6
2.5

0.5
2
0.4
1.5
0.3

1
0.2

0.5
0.1

0 0

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Satisfaction Satisfaction
 DAS/MAT 9
Agreement on: WAYS OF DEALING WITH PARENTS OR IN-
LAWS

1 - Always Disagree
1 4.5
.

2 - Almost Always Disagree

0.9
3 - Frequently Disagree 4
Probability of each response

.
0.8 4 - Occasionally Disagree 3.5

0.7
5 - Almost Always Agree
3
6 - Always Agree

Information (theta)
0.6
2.5
0.5
2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Satisfaction Satisfaction
 MAT 12
In leisure time, do you (and does your mate) prefer to be “on the
go” or to stay at home?

1 1 - Mismatch 4.5
.

0.9 2 - Both on-the-go 4

.
3 - Both stay-at-home
Probability of each response

0.8
3.5

0.7
3

Information (theta)
0.6
2.5

0.5
2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Satisfaction
Satisfaction
From Items to Scales
 A scale’s information

= sum of information from each item

 How informative

Across different levels of happiness

Scale Information
60

DAS (32)
50
QMI (6)
Scale Information

RAS (7)
40
MAT (16)

30 DAS-4 (4)

20

10

0
-3 -2 -1 0 1 2 3

Satisfaction
Summary
 MAT and DAS have poor items

 Increases NOISE

 MAT-15 no better than 4-item scale

 DAS-32 little better than 6-item scale

 Assess satisfaction, but not very efficiently

 Poor thermometers
Creating the CSI
 141 item pool

 Screen for contaminating items

 Screen for redundant items

 IRT on remaining 66 items

 Select 32 most effective

 Parameter Invariance
RANDOM SAMPLE HALVES MALE vs. FEMALE

5 7

.
6
.

r = 0.998 r = .991 5

b's est in FEMALE respondents

3
b's est in odd-row subjects

2 3

2
1

0
0
-5 -4 -3 -2 -1 0 1 2 3 4 5
-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7
-1
-1

-2
-2
-3

-3 -4

-5
-4

-6

-5

b's est in even-row subjects -7

b's est in MALE respondents

Basic Psychometrics

Correlations
Distress
Alpha
Cut Score 1 2 3 4 5 6

1. DAS .94 97.5 -- -- -- -- -- --

2. MAT .84 95.5 .90 -- -- -- -- --

3. QMI .96 24.5 .85 .87 -- -- -- --

4. RAS .92 23.5 .86 .87 .91 -- -- --

5. CSI-32 .98 104.5 .91 .91 .94 .96 -- --

6. CSI-16 .98 51.5 .89 .90 .96 .95 .98 --

7. CSI-4 .94 13.5 .87 .88 .93 .94 .97 .97

Correlations with Anchors
Thoughts of Positive Hostile Sexual
Stress Neuroticism
Breakup Communication Conflict Chemistry

DAS -.74 .73 -.53 -.54 .42 -.40

MAT -.74 .69 -.49 -.49 .41 -.38

CSI-32 -.78 .71 -.52 -.48 .45 -.38

CSI-16 -.78 .71 -.53 -.49 .43 -.38

CSI-4 -.75 .69 -.52 -.47 .41 -.36

Criterion Validity
 DAS Distress groups
 Current gold-standard

 DAS score < 97.5

 1027 DAS distressed P’s

 ROC’s to identify CSI cut scores

 Identified CSI distressed P’s

 91% agreement w/ DAS

Summary
 Operate similar across
 Male vs. Female
 Older vs. Younger
 Married vs. Engaged vs. Dating

 CSI measures same construct

 Nearly identical correlations
 Highly similar screen for distress

 Evaluating Possible Improvement

 CSI-32 vs. DAS-32
 CSI-16 vs. MAT-15
 CSI-4 vs. DAS-4

 More information?
 Less noise?
 Better thermometer?
Scale Information
60 CSI-32
CSI-16
50 CSI-4
DAS-32
Information

40
MAT-15
DAS-4
30

20

10

0
-3 -2 -1 0 1 2 3
Satisfaction
Relative Efficacies
7 CSI-16 vs MAT
CSI-32 vs DAS
6
CSI-4 vs DAS-4
5
Effective Length

0
-3 -2 -1 0 1 2 3
Satisfaction (SD's)
Satisfaction Groups
 IRT satisfaction estimates
 For each subject
 Based on MAT, DAS, & CSI items

 (equivalent of SAT scores)

 Created satisfaction groups

 N = 265 in each group
 Levels of sat. HIGHLY similar within each group

 MAT, DAS & CSI scores also similar?

Precision: CSI-32 vs. DAS
Precision: CSI-16 vs. MAT
Effect Size
 Ability to detect difference
 Between groups
 Pre – Post

 Effect Size = M1 – M2 .
pooled SD

 Difference in SD units

 Power for detecting ’s in SAT groups

Power: CSI-32 vs. DAS
2.5
.

DAS
Effect Sizes (Cohen's d)

2 CSI(32)

1.5

0.5

0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Adjacent Satisfaction Group Contrasts

Power: CSI-16 vs. MAT
2.5
.

MAT
2 CSI(16)
Effect Sizes (Cohen's d)

1.5

0.5

0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Adjacent Satisfaction Group Contrasts

Conclusions
 CSI scales
 More information
 Less noise
 More power

 Better thermometers

 NEXT STEP
 True over time?
 Better at detecting change?
Studies 2, 3, 4: Method
 Study 2
 596 online respondents
 1 and 2 week follow ups (n = 267)
 CSI, MAT, DAS

 Study 3
 398 online respondents
 1, 2, 3, 4, 6 and 12 mo follow ups (n = 156)
 CSI, MAT, DAS

 Study 4
 1,062 online respondents
 1, 2, 3, 4, 6 and 12 mo follow ups (n = 545)
 CSI, MAT
Studies 2-4: Demographics
 SAMPLE
 N = 2,056 initial respondents
 N = 968 (47%) respondents with longitudinal data

 AGE
 M = 27.7yo (9.3yrs)

 GENDER
 71% Female
 29% Male

 RACE
 83% Caucasian
 5% Asian
 4% African American
 4% Latino

 SES
 10% High school diploma or less
 25K avg yearly income
Studies 2-4: Relationships
 Relationship Types
 37% Married: 7.9 yrs (7.9 yrs)
 13% Engaged: 3.2 yrs (2.4 yrs)
 50% Dating: 1.8 yrs (1.9 yrs)

 Relationship Satisfaction (MAT)

 Married: 108 (32)
 Engaged: 122 (24)
 Dating: 116 (24)

 Dissatisfied Respondents
 24% (n = 487)
Change Criterion
Much Somewhat A little Stayed the A little Somewhat Much
WORSE WORSE WORSE SAME BETTER BETTER BETTER

-3 -2 -1 0 +1 +2 +3

 How much has each of these changed?

 Overall happiness in the relationship
 Feeling close and connected
 Stability of the relationship

 Averaged responses
 Alpha = .92

 Agree with MAT, DAS, & CSI scores?

Noise over time (SERM)
 Score scatter in “no change” group
Much Somewhat A little Stayed A little Somewha Much
WORSE WORSE WORSE the SAME BETTER t BETTER BETTER

-3 -2 -1 0 +1 +2 +3

 238 “no change” at 1st assessment

 Repeated Measures MANOVA

 Scatter (noise) in scale scores across time

 SERM = 2*MSE
Detecting Individual Change
 Can we detect individual change?
 Minimal Detectible Change (MDC95)
• RCI: Jacobson & Truax (1991)
• MDC95: Stratford et al. (1996)

 Pre-Post score change

• In one individual
• Necessary to exceed noise

 MDC95 (SD units) = 1.96*SERM .

SD
Minimum Detectible Change
How much must an individual’s score
shift to show significant change?

0.8 C*
C
0.7 C*
MDC's in SD Units

0.6 B
A
0.5
0.4
0.3
0.2
0.1
0
CSI-32 CSI-16 CSI-4 DAS MAT
Detecting Individual Change
 CSI scales more sensitive
 Required smaller pre-post score shifts

 Longer scales more sensitive

 CSI-32 > CSI-16 > CSI-4

 MAT & DAS not as sensitive

 Operated no better than CSI-4
Detecting Group Differences
 Can we detect clinically distinct groups?
 Improved vs. No-change
 Deteriorated vs. No-change

 Minimal Clinically Important Difference (MCID)

• Guyatt, Walter & Norman (1987)

 MCID Effect Size = M(improved) – M(no change)

Noise over time (SERM)

 HLM framework
• Global change predicting  scores on scales
• 2,475 points of change from 968 respondents

• Improved vs. Deteriorated

• Satisfied vs. Dissatisfied
• Gender effects
 MCID Effect Sizes
How well can we detect naturally
occurring change?
A
3.0 B
1 Point of Global Change

CSI-32
Effect Size to Detect

2.5 C
CSI-16
2.0
CSI-4
D DAS
A A
1.5 B MAT
E
A B
1.0 C
D C C C
0.5
A A B
B B
0.0
Deterioration Improvement Deterioration Improvement
Dissatisfied Respondents Satisfied Respondents
Differences by Gender
 Scales showed slightly smaller effect sizes in men

0.1
Reduction in Effect Sizes

-0.1
*
-0.2 *
* *
-0.3

-0.4

-0.5
CSI-32 CSI-16 CSI-4 DAS MAT
Detecting Group Differences
 CSI-32 & CSI-16
 Out performed DAS & MAT
• Improvement / Deterioration
• Satisfied / Dissatisfied

 CSI-4
 Deterioration: Out performed DAS & MAT
 Improvement: Equivalent to DAS & MAT

 Weak gender effect

 All scales slightly less responsive in males
Summary of Development
 CSI scales represent improved thermometers
 Developed with IRT / FA
 No contaminating items
 Non-redundant items
 Most informative items

 Still measure satisfaction

 Consistent with MAT / DAS

 Offer greater power

 More information
 Less noise

 More sensitive cross-sectionally

 Detecting group differences

 More responsive over time

 Detecting change in a single individual
 Detecting differences between clinical groups
PART 2: Administration
 See CSI handout

 Spouses complete separately

 No discussion during administration
 Want unique perspectives

 Inform of confidentiality limits

 Feedback given?
 Dyadic or individual feedback?
 Normative data

 Should take 3-4 minutes

Scoring
 See CSI scoring handout

 Sum the item responses

 10 reverse scored items
• High sat options offered first (items 2-5)
• Reversed wording (items 10, 15…)

 Total scores
 Range from 0-161
Interpretation
 Box Plots

Lowest Median Highest

25% of 25% of
scores 2nd quartile 3rd quartile scores
of scores of scores

 Dissatisfaction Cut Score

 Scores below 104.5
Norms in Dating Individuals

N = 1477

N = 2191

N = 415
Norms in Engaged Individuals

N = 551

N = 141
Norms in Married Individuals

N = 1129

N = 735
Norms in Married Individuals

N = 271

N = 321