Biofam Three-Domain Family Life Dataset

The biofam dataset is originally a single-channel sequence dataset. In the original version, each person has one family-life sequence from age 15 to age 30. Each yearly state combines several aspects of family life into one code, such as whether the person lives with their parents, has left home, is married, has children, or is divorced.

For multidomain sequence analysis, we also provide a three-domain version of the biofam dataset. Instead of storing family life as one combined state, this version separates family life into three simpler domains:

1. marriage domain
2. residence / leaving-home domain
3. parenthood / child domain

This structure follows the multichannel representation used in the seqHMM package (Helske and Helske, 2019), where the original biofam data are split into three parallel life domains: marriage, parenthood, and residence. In seqHMM, the three channels are used to represent whether someone is single/married/divorced, childless/has children, and living with parents/left home.

Why do we split biofam into three domains?

The original biofam states are combined states.

For example, in the original biofam dataset:

Code	State
0	Parent
1	Left
2	Married
3	Left+Marr
4	Child
5	Left+Child
6	Left+Marr+Child
7	Divorced

A state such as Left+Marr+Child contains three pieces of information at the same time:

Left+Marr+Child
= left parental home
+ married
+ has children

This is compact, but it can be difficult for beginners to see which part of the trajectory comes from which life domain.

The three-domain version makes this easier. It separates the same information into three parallel sequences:

Residence domain:   whether the person has left home
Marriage domain:    whether the person is married
Child domain:       whether the person has children

So instead of reading one combined state, we can read three simpler sequences side by side.

Files included

We provide three domain-specific files:

File	Domain	Meaning
biofam_married_domain.csv	Marriage domain	Whether the person is married
biofam_left_domain.csv	Residence domain	Whether the person has left the parental home
biofam_child_domain.csv	Parenthood domain	Whether the person has children

Each file has the same basic structure:

id, 15, 16, 17, ..., 30

The id column identifies the person. The columns 15 to 30 represent ages 15 to 30. Each cell records the person’s state in that domain at that age.

Important idea: one person, three parallel sequences

The same person appears in all three files with the same id.

For example, person 1 has one sequence in the marriage domain, one sequence in the residence domain, and one sequence in the child domain.

Together, these three sequences describe the person’s family-life trajectory.

Person 1
Marriage domain:   not married → married
Residence domain:  with parents → left home
Child domain:      no children → has children

This is why the dataset is called multidomain or multichannel sequence data: each person has several parallel sequences observed over the same ages.

Marriage domain

File: biofam_married_domain.csv

This file records whether each person is married at each age.

State coding

Code	Meaning
0	Not married
1	Married

Example

id,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30
1,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1
2,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1
3,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1
4,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
5,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1

For person 1, the sequence is:

0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1

This means that the person was not married from age 15 to age 23, and was married from age 24 onward.

In a more readable form:

Not married → Not married → ... → Married → Married

Residence / leaving-home domain

File: biofam_left_domain.csv

This file records whether each person has left the parental home.

State coding

Code	Meaning
0	Living with parents
1	Left parental home

Example

id,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30
1,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1
2,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
3,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1
4,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1
5,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1

For person 2, the sequence is:

0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1

This means that the person was living with their parents at age 15, but had left the parental home by age 16 and remained in the left home state afterwards.

In a more readable form:

With parents → Left home → Left home → ... → Left home

Child / parenthood domain

File: biofam_child_domain.csv

This file records whether each person has children.

The file name uses child for simplicity. Substantively, this is the parenthood domain: it tells us whether the person is childless or has children at each age.

State coding

Code	Meaning
0	No children
1	Has children

Example

id,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30
1,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1
2,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1
3,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1
4,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
5,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1

For person 1, the sequence is:

0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1

This means that the person had no children from age 15 to age 24, and had children from age 25 onward.

In a more readable form:

No children → No children → ... → Has children → Has children

Reading the three domains together

The real value of the three-domain version is that the three files can be read together.

For example, suppose person 1 has the following three sequences:

Marriage domain:
0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1

Residence domain:
0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1

Child domain:
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1

This can be interpreted as:

Age 15–23:
not married, living with parents, no children

Age 24:
married, left parental home, no children

Age 25–30:
married, left parental home, has children

So the person first remained in the parental home and was not married, then left home and got married around age 24, and then had children around age 25.

This example shows why multidomain sequence data are useful. They allow us to see not only whether life events happened, but also how different life domains changed together over time.

Relationship to the original biofam states

The three-domain files are derived from the original combined biofam states.

The logic is simple: each original state is decomposed into separate pieces of information.

Original biofam state	Marriage domain	Residence domain	Child domain
Parent	Not married	With parents	No children
Left	Not married	Left home	No children
Married	Married	With parents	No children
Left+Marr	Married	Left home	No children
Child	Not married	With parents	Has children
Left+Child	Not married	Left home	Has children
Left+Marr+Child	Married	Left home	Has children
Divorced	Not married / divorced depending on coding choice	Usually left home	May depend on the original coding

In our simplified three-domain files, each domain is represented as a binary sequence using 0 and 1.

This makes the files easy to understand and convenient for teaching multidomain sequence analysis.

Difference between single-channel and three-domain data

The original biofam dataset stores family life as one combined sequence.

For example:

Parent → Parent → Left → Left+Marr → Left+Marr+Child

The three-domain version stores the same kind of information as three parallel sequences:

Residence:
With parents → With parents → Left home → Left home → Left home

Marriage:
Not married → Not married → Not married → Married → Married

Child:
No children → No children → No children → No children → Has children

Both formats describe family life courses. The difference is in how the information is represented.

The single-channel version is compact because all dimensions are combined into one state. The three-domain version is more explicit because each life domain has its own sequence.

Why is this useful?

The three-domain version is useful when we want to study several life-course dimensions at the same time.

It can help us ask questions such as:

• Do people usually leave home before marriage?
• How often do marriage and leaving home happen around the same age?
• Do people tend to have children soon after marriage?
• Are there different patterns of family formation across men and women?
• Do birth cohorts differ in the timing of leaving home, marriage, and parenthood?
• Are some trajectories more synchronized across domains than others?

These questions are difficult to answer if we only look at one final outcome. They require us to observe the order and timing of changes across several domains.

File structure

Each domain file has the same structure:

Column	Description
id	Individual identifier
15	State at age 15
16	State at age 16
17	State at age 17
...	Yearly states
30	State at age 30

The sequence length is 16 because the data cover ages 15 to 30.

Summary

The three-domain biofam dataset is a multidomain version of the original biofam family-life sequence data.

Instead of representing family life as one combined state sequence, it separates family life into three parallel domains:

Marriage domain:    not married / married
Residence domain:   with parents / left home
Child domain:       no children / has children

This representation is especially helpful for beginners because it makes the structure of family-life trajectories easier to see. It is also useful for multidomain sequence analysis, where the goal is to study how several life domains unfold together over time.

Reference

Müller, N. S., Studer, M., & Ritschard, G. (2007). Classification de parcours de vie à l'aide de l'optimal matching. In XIVe Rencontre de la Société francophone de classification (SFC 2007), Paris, 5–7 September 2007, pp. 157–160.

Helske, S., & Helske, J. (2019). Mixture Hidden Markov Models for Sequence Data: The seqHMM Package in R. Journal of Statistical Software.

TraMineR documentation: biofam, Example data set: Family life states from the Swiss Household Panel biographical survey.

Author: Yuqi Liang