1 Family Variables: Latent Class Latent Transtion Alan C. Acock Presented at the Conference on...

Preview:

DESCRIPTION

3 A family variable Individuals characteristics & actions may create family variables These are family variables because they locate the opportunity structure of all family members, not just the individual A child with a profound disability A parent who is incarcerated

Citation preview

1

Family Variables: Latent Class & Latent Transtion

Alan C. AcockPresented at the Conference on Research with

Dyads and familiesPurdue University

May, 2010

2

Family Variables• Demographic variables are often

considered family variables• Household income• Ethnicity• Location

• Family variables because they influence opportunity structure of all family members

3

A family variable• Individuals characteristics & actions

may create family variables• These are family variables because

they locate the opportunity structure of all family members, not just the individual• A child with a profound disability• A parent who is incarcerated

4

From Variable to Person to Family Centered

Analysis• Most research is variable centered a

single dependent variable• Marital happiness• What other variables predict

• Number of children• Spouse’s social support

5

From Variable to Person to Family Centered

Analysis• Variable Centered research can

be extended• Multiple outcomes, Longitudinal

Analysis• Including family level variables• Variables about other family members

• This is still variable centered

6

From Variable to Person to Family Centered

Analysis• A variable’s distribution may belie

the fact that there are multiple distributions of the variable

• Person centered analysis identifies clusters of people who have fundamentally different distributions

7

From Variable to Person to Family Centered

Analysis• A high school may have a distribution on a

standardized test• Person centered research says there may

be two or more distributions• What explains the distribution in which a

person is located?• What outcomes flow from the distribution

in which a person is located?

8

From Variable to Person to Family Centered

Analysis

9

From Variable to Person to Family Centered

Analysis

TextTextTextText

Schaeffer, C.M., Petras, H., Ialongo, N., Poduska, J. & Kellam, S. (2003). Modeling growth in boys aggressive behavior across elementary school: Links to later criminal involvement, conduct disorder, and antisocial personality disorder. Developmental Psychology, 39, 1020-1035.

10

From Variable to Person to Family Centered Analysis

• Individual Level Variable: An Attitude• Multiple indicators • May have multiple informants

11

From Variable to Person to Family Centered Analysis

12

Latent Class Analysis of Family Level Variables• Identify Subsets of families who

have fundamentally different distributions on variables

• Latent Classes are homogeneous within clusters

• Latent Classes are heterogeneous between clusters

13

Latent Class Analysis of Family Level Variables

14

Latent Class Analysis of Family Level Variables

15

Latent Class Analysis of Family Level Variables• We may want to know what causes

a families to be in one class rather than another

• We may want to know what are the consequences of class membership for individual members as well as the family

16

Latent Class Analysis of Family Level Variables

17

Mixture Models & Longitudinal Analysis

• LCA is a special type of Mixture modeling

• Mixture modeling identifies latent classes in any type of analysis

• There are two broad classes of mixture modeling for longitudinal analysis

18

Classes Applied to a Variable Centered

Approach• We construct a growth curve• Then will see if there are two or

more latent classes that• Are heterogeneous across classes• Have homogeneous growth

trajectories within classes

19

Classes Applied to aVariable Centered

Approach

20

An Auto-Regressive Model: Latent Transition

Analysis• LTA begins with latent classes• Identifies how these classes are

stable and how they change over time

• Focus is on transitions over time

21

An Auto-Regressive Model: Latent Transition

Analysis

22

LTA with Antecedent Covariates & Distal Outcomes

23

LTA with Mover-Stayer Latent Transition Model

24

Summary

• Family variables are scores attached to families that influence any and all family members

• We can treat family members, e.g., mother, father, focal child as multiple informants

25

Summary

• Mixture Models identify clusters of units being analyszed that are homogeneous within clusters and heterogeneous between clusters

• Latent class analysi (LCA) identifies these clusters of families or whatever unit of analysis is used

26

Summary

• Clusters may be ordinal such as strong moderate and challenged

• Clusters may be dis-ordinal where one cluster has different strengths than another cluster

27

Summary

• Research should identify sets of covariates that lead to membership in different clusters

• Research should identify sets of outcomes that are caused by cluster membership

28

Summary• Latent transition analysis is an

alternative to growth models as a way of studying change

• LTA focuses on transitions between classes.

• Why do some families move toward increased solidarity when their oldest child becomes an adolescent?

29

Summary

• Research should identify sets of covariates that differentiate what class a family is in

• Research should identify sets of outcomes that very with the class the family is in

30

Summary

• The mover-stayer model identifies classes of people who are moved and those who are not.

Recommended