Within families: family-wide and child-specific influences on childrens socio-emotional development Jennifer Jenkins, Jon Rasbash, Tom OConnor.

Within families: family-wide and child-specific influences on children’s socio-emotional

development

Jennifer Jenkins, Jon Rasbash, Tom O’Connor

Behavioral genetic findings of siblings being so different from one another once genetic effects

were controlled

What is the effect of the family environment?

Focus on the similar and different experiences

of siblings in families

Themes• Do family-wide or child-specific aspects of

the environment predict change in child behavior?

• How similar are children’s experiences in families? Does this vary as a function of stresses in the environment?

• What are children’s own contributions to the stressful environments that they experience?

Datasets• NLSCY, ABSS, NEAD• All involve the inclusion of multiple

children per family: between 2-4 depending on dataset

• Some involve multi-informant data• Some results involve examining change in

the response variable: longitudinal design• All use multilevel modeling for the analysis

Jenkins, Rasbash, O’Connor (2003) DPJenkins, Simpson, Dunn, Rasbash, O’Connor (2005) CDJenkins, Dunn, O’Connor, Rasbash, Behnke, JFP. In pressRasbash, Jenkins, O’Connor, In preparation

OutcomeVariable

Between FamilyComparisons

Family A Family B

1 2

FamilyLevel

Child Level

Majority of environmental studies of family influencesfamily and child-specific processes are confounded

Environmental studies using sibling design: unconfounds family andchild

Family A

Family B

Family Level

Child Level

Between family comparisons

Within family comparisons

1 2 3

Cross-level interactions

Measures at the family and child-specific levels

3

FamilyA

FamilyB

Family averagenegativity

Child’sdeviation fromthe family mean

1 2

Lowincome/education

Child’sfriendshipsoutside family

2

Do family-wide or child-specific aspects of the environment predict

change in child behavior?

Illustrate with results from sibling study

Are there within family differences on sibling dyad

negativity?

0

20

40

60

80

100

120

Between familyvarianceWithin familyvariance

ICC = .52

Shared effects: Av maternal negativity predicts an increase in sibling negativity over 2 years

0

4

8

Av maternal negativity one unit change inmaternal negativity

Family average is a stronger predictor of sibling negativity than dyad-specific negativity

Mean of sibling negativity as a function of gender of sibling

dyad

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6.8

7

Girl dyadBoy dyadMixed dyad

Girl dyads differ significantly from mixed dyads

*

*

Thus there are some systematic reasons that some dyads get on

better than others

And shared experiences are important

To what extent do siblings live in shared environments?

Shared family environments?

Exposure to parental conflict

Family Blue Family Yellow Family Pink

Families differ from one another on how much parental conflict children experience

Shared family environments?

Children within families differ from one another on how much parental conflict they experience

Exposure to parental conflict

Family Blue Family Yellow Family Pink

Sibling similarity on experiences surrounding parental conflict

0

0.2

0.4

0.6

argument about children exposure to conflict

ICC

To what extent do siblings live in shared

environments? Does this vary as a function of environmental stress?

Modeling differential experience in families

Differential parental positivity as a function of SES, marital problems and family size.

-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

household ses

1

2

3

4

5

diffe

rentia

l po

sitive p

are

nting

family size = 2, no marital problemsfamily size = 2, marital problemsfamily size > 2, marital problemsfamily size > 2, no marital problems

Differential parental hostility as a function of single parenthood and marital dissatisfaction.

Intact/no maritalproblem

Marital problem Single parent4.2

5.2

6.2

7.2

differential negativity

Limitations of this method

Measurement problems. Although in some of the studies the IV and DV are based on different informants, the family clustering information is based on single informant. Degree of family clustering that we see may be related to same person reporting on measures for different siblings

Another method for examining similar and differential

experiences in families as well as consistency of behavior when

interacting with different members of the family

Social relations model

• Every person in a family interacts with every other person• Rate each person’s expression of negativity and positivity

towards every other family member• Data are observational

Within family structure

Family 1…

Relationship: c1c2 c1m c1f c2c1 c2m c2f mc1 mc2 mf fc1 fc2 fm

Actor: c1 c2 m f

Partner: c1 c2 m f

Dyad d1 d2 d3 d4 d5 d6

We start with 12 relationship scores in each family. These can be classified :

partner dyad and familyactor

Interpretation of variance components

Family:the extent to which families differ from one another

Actor: the extent to which individuals act similarly across relationships.

Partner: the extent to which individuals elicit the same behaviors from others.

Dyad: The extent to which emotion expression is specific to the dyad and based on reciprocity.

Consistency of individual emotional expression across

dyads: a function of the dyad or the individual?

% of variance explained by actor, partner and dyad for positivity and

negativity:

05

1015202530354045

Actor Partner Dyad

Positivity Negativity

• People are very consistent in their positivity across relationships irrespective of what their interactional partner is doing. Not so of negativity.

• The partner effect is surprisingly small

Negativity is less internal to the person and more sparked off by behavior of interactional partner

than positivity

High reciprocity for negativity

Extent to which families are different from one another

% of variance at the family level

0

2

4

6

8

10

12

14

16

18

20

Positivity Negativity

Do characteristics of individuals affect the dyads in which they interact or do they exert their

influence across the whole family?

Example depression

An individual’s depression score may explain family level

variance or only variance in dyads in which the individual is a

member.

Example of data coding to examine effects on families or dyads

Family (spillover) and dyad effects of depression on NEGATIVITY

Depression score Significance

Child 1 at family level *

Child 2 at family level *

Mother at family level ns

Father at family level *

Child 1 at dyad level ns

Child 2 at dyad level ns

Mother at dyad level *

Father at dyad level ns

Family (spillover) and dyad effects of depression on POSITIVITY

Depression score Significance

C1 at family level ns

C2 at family level ns

Mother at family level ns

Father at family level *

C1 at dyad level ns

C2 at dyad level ns

Mother at dyad level ns

Father at dyad level ns

Advantages of multilevel methods for family data

• Differentiating between family-wide and child-specific processes in families

• Highly stressful environments increase the variation in the within family environment

• Social relations model allows us to look at consistency of individual’s behavior in families, how different families are from one another and why

Disadvantages of multilevel approach to families

),0(~),0(~

),0(~),0(~),0(~2

),,(2

222

),,(0),,(

emlkjidlm

pkmajmfm

mlkjilmkmjmmmlkji

NeNd

NpNaNf

edpafy

For psychologists the statistics get very

complicated

Cross discipline collaboration is essential!!!