Who Counts as "Family"? Classic Practices for Defining Household Members in Large Surveys

A household interview often begins not with opening the questionnaire, but with settling the roster of who belongs to this household. Get the roster wrong, and everything that follows — household income, per-capita spending, household size — goes wrong too. This article systematically reviews how major surveys in China and abroad define who counts as a household member.

Why "Who Counts as Family" Is the Bedrock of Interviewing

Household surveys are organized around the household unit. Whatever the questionnaire asks, the household itself — who belongs to it, how many people it contains, and how they are related — is the denominator and the coordinate system for all the data.

• If the roster omits a son working away from home, household income is underestimated and average household size shrinks;
• If the roster counts a relative staying over for a short visit, per-capita indicators are distorted and consumption data are diluted;
• If relationships are not clarified, everything downstream — "relationship to household head" and the generational structure — collapses.

That is why, without exception, major surveys in China and abroad write member-definition rules into hard protocols and interviewer manuals, rather than leaving them to an interviewer's on-the-spot judgment. This is exactly the dividing line between professional and amateur interviewing.

First, Untangle Three Concepts: Family ≠ Household ≠ Family Household

To talk about "who counts as family," we first need to separate three terms that are often used interchangeably:

The international distinction is clean: family asks "what is the relationship," while household asks "how do people live and get by." A person can belong to one family yet live in another household — a child working away from home, for example, is still "family" but does not currently "live under the same roof."

China's population censuses divide households into family households and collective households. The former are composed mainly of people related as family members who live and share life together (a single person living alone also forms a separate household). The latter are groups of unrelated people who live together collectively — such as student dormitories or workplace dormitories — counted as one household per housing unit. The classification itself reveals how the two main threads, relationship and residence, intersect.

International Experience: Two Frameworks and One Main Thread

1. United Nations: de facto and de jure

The United Nations Principles and Recommendations for Population and Housing Censuses sets out two basic frameworks:

• De facto population: people actually present at the time of the survey = usual residents who are present + visitors/transients;
• De jure (usual-residence) population: people who habitually live in a place = usual residents who are present + those temporarily away.

To determine the "place of usual residence," UNECE and Eurostat recommend a 12-month time criterion — having lived there continuously for most of the past 12 months (at least 6 months and 1 day), or intending to live there for at least 6 months.

The most important principle is this: each person should have one — and only one — place of usual residence (one and only one place of usual residence). This directly serves the fundamental goal of statistics: no double counting and no omission.

2. US Census Bureau: the "Usual Residence" Rule

The US Census Bureau's definition of "usual residence" is simple enough to memorize: the place where a person lives and sleeps most of the time (or where they consider themselves to "usually reside"). On this basis, a household roster is determined by two principles:

• People who usually live here, whether present at the time of the survey or temporarily away (on business, traveling, hospitalized, or a newborn still in hospital);
• People living here who have no usual residence elsewhere.

At the same time it excludes: people whose usual residence is elsewhere, people in long-term institutional custody (prisons, psychiatric hospitals, nursing homes), college students whose usual residence is elsewhere, and military personnel whose usual residence is elsewhere.

Interestingly, even within the same Census Bureau, different surveys use different frameworks. The CPS (Current Population Survey) uses usual residence, while the ACS (American Community Survey) uses "current residence" — anyone who has lived at an address for at least 2 months is counted, in order to better reflect an area's present characteristics. The lesson: a framework is never absolutely right or wrong, only more or less suited to the research purpose — but it must be fixed in advance and applied consistently throughout.

3. World Bank LSMS: "Eating from the Same Pot"

The World Bank's Living Standards Measurement Study (LSMS) has long defined a household in one highly down-to-earth phrase: a group of people who live together and share meals ("eat from the same pot"). The logic is pragmatic — poverty and consumption measurement needs a unit that shares economic resources, not a legal or household-registration unit.

4. European ESS / US GSS: Sample the Household First, Then the Person

Surveys centered on individual attitudes, such as the European Social Survey (ESS) and the US General Social Survey (GSS), likewise first delimit households using the usual-residence rule plus the roster method, and then randomly select a respondent within the household. Member definition is still the first step — the difference is that they care about "which household to sample from" rather than a complete portrait of the household.

Summary: The main thread of international experience is to center on "usual residence," build the roster with the relationship-to-household-head method, and hold the boundary with shared economics and living together.

China's Experience: From Census to Academic Surveys

1. China's 7th National Population Census: "Usual-Resident" Framework plus Household Enumeration

The 7th census registers by household, following the principle of enumeration by current place of residence, and defines the "usual-resident population" as the sum of four categories:

• People living in this township/town/subdistrict whose household registration is also in this township/town/subdistrict;
• People living in this township/town/subdistrict whose household registration is elsewhere and who have been away from their registered place for more than six months;
• People whose household registration is pending;
• People who originally lived in this township/town/subdistrict and are now working or studying in Hong Kong, Macao, Taiwan, or abroad.

For "who should be registered in this household," the 7th census gives very specific, ready-to-copy rules:

• Should be registered: people who spent the survey reference night in this household; people who usually live here but were away that night on temporary business, family visits, travel, or night shift; children in full-time nursery care and primary/secondary boarding students; and other people whose household registration is at this address;
• Should not be registered: active-duty military personnel and armed police; and people temporarily staying in this household that night for temporary reasons.

The "relationship to household head" item covers: household head, spouse, children, parents, parents-in-law, grandparents, daughters/sons-in-law, grandchildren, siblings, and others — a standardized relationship checklist.

2. CFPS (China Family Panel Studies): Kinship plus Economic Ties ("Sharing the Same Stove")

Peking University's CFPS is the benchmark for family surveys in China, and its definition of "family member" is especially worth learning from for interviewing practitioners. It does not treat "living together" as a necessary condition, but instead uses two criteria:

• Kinship: members must be related by marriage, blood, or adoption (a nanny or driver without such a relationship is not a member, even if they live there long-term);
• Economic ties ("sharing the same stove"): the person must "depend on the family for support" or "need to support the family."

The key conclusion: whether someone lives under the same roof does not determine whether they are a family member. A child working away from home who regularly sends money back is still a family member, while an adult child who has started a family and is fully financially independent is not.

The CFPS interviewer manual also specifically reminds respondents about seven categories of people who are "not at home but may have economic ties": studying away, working away, living as clergy, visiting relatives or friends, serving a prison sentence, in military service, or abroad. This reminder list is a classic design in interviewing practice.

3. CSS, CGSS, CHFS, CHARLS: Different Roads, Same Destination

The Chinese Social Survey (CSS), Chinese General Social Survey (CGSS), China Household Finance Survey (CHFS), and China Health and Retirement Longitudinal Study (CHARLS) phrase their definitions of "family/household" differently, but their core is the same: a combination of three elements — blood/marriage/adoption ties, living together, and economic sharing.

CHFS emphasizes "living together, sharing life together, and jointly bearing economic obligations." CHARLS draws the household boundary as "living in the same housing unit + having kinship ties + sharing economic resources."

In a word: China's major surveys converge strongly on "who counts as family." Relationship and economic ties are the two hard bars; whether people live together is a variable that can flex.

Summary of Classic Practices: A Reusable Decision Framework

Distilling the experience at home and abroad yields a four-step decision framework for interviewing:

1. Step 1: First fix the "household head / reference person." International practice (the US census and most academic surveys) always identifies a "reference person" or "household head" first, and the roster is built around them. The 7th census stipulates that the head of a family household is registered first.
2. Step 2: Build the roster person by person using "relationship to the household head" (roster method). Starting from the head and probing level by level — spouse → children → parents → … — misses far fewer people than the open-ended question "who else is in this household?"
3. Step 3: Fall back on the "usual-residence principle," with a time threshold of 6 or 12 months. For "does this person actually belong to this household," use "usual residence + time threshold" to decide objectively: someone away from their registered/usual residence for more than six months is counted elsewhere, while a short-term stay below the threshold does not count.
4. Step 4: Hold the special boundary of the Chinese family with "economic ties / eating and living together." For people who are related but do not live together (migrant workers, children at boarding school), ask the key question: "Does he or she currently depend on the family for support, or send money home?" This is the most practical single question for deciding whether someone still counts as family.

How to Decide Six Types of "Difficult Cases" (an Interviewer's Decision Checklist)

Four Practical Tips for Interviewers

• Draw the relationship map before filling in the roster. As you ask about "relationship to the household head," sketch a generational diagram at the same time — this keeps omissions to a minimum.
• Double-confirm with "time + money." For each borderline person, pin them down with two questions: How long have they lived here? Who supports whom?
• Handle sensitive situations gently. For imprisonment, monastic life, separation, adoption, and similar cases, use neutral phrasing such as "is there anyone whose living situation has recently changed?" to avoid causing offense.
• Record your criteria and leave a trail for review. Jot down what standard each household used and who was excluded — this becomes evidence during review and follow-up.

Conclusion

"Who counts as family" may look like a single phrase, but it is actually the first gate for interviewing quality. The consensus at home and abroad is highly consistent: use relationship to fix the structure, usual residence to fix membership, economic ties to hold the boundary, and a time standard to make the final call. Write these four rules into the manual and drill them into interviewers' hands, and the data quality of household interviews wins at the starting line.

This article was compiled and written by Beijing Tongtai Network Technology Co., Ltd. We focus on developing systems for large-scale face-to-face surveys and have long undertaken face-to-face survey system projects for households, enterprises, and institutions. You are welcome to discuss cooperation with us.

References

• China's 7th National Population Census: principles for registering and determining the usual-resident population
• CFPS Mini-Lesson: The Definition of Family Members (Institute of Social Science Survey, Peking University)
• CFPS 2010 User Manual (defining household and family members)
• US Census Bureau: Group Quarters and Residence Rules for Income
• UNECE / Eurostat: Concepts and Definitions for the 2010 Population and Housing Census
• Multilingual Demographic Dictionary: de facto / de jure population