897. The demographic characteristics of sex, age and marital status are core variables that are fundamental in classifying and reporting on other information from the census to foster understanding of socioeconomic and demographic issues. In the case of sex and age, it is essential to record this information for every person for whom census information is collected or compiled. If this information is incomplete, it should therefore be derived or imputed for census purposes rather than being reported as “not stated”. In the case of marital status, as discussed below in paragraph ‎922, it is recommended to collect this information at least for persons aged 15 years and over.

898. A population census may provide an opportunity to collect the data needed for estimating fertility at both national and sub-national levels. The use of a population census as a vehicle for producing fertility estimates is particularly important in countries lacking a reliable registration system for the production of vital statistics, because of the opportunity that census data can provide for estimating vital rates that would not otherwise be available. Even in countries with complete birth registration, however, some demographic topics – such as numbers of children born alive, children still living, and age at marriage or union – are still of great value as they can be cross-tabulated with socioeconomic or other characteristics collected in the census, which would not be possible with civil and vital registration data alone. For example, the census allows collecting data on children born to immigrants before their arrival to the country, which are not captured in the civil registration system.

899. Sex is designated as a core topic and relates to biological attributes that result in one being labelled as male or female. Disaggregating the other demographic, social and economic characteristics by sex is important, as is the use of sex data for survey controls. Some countries are moving from collecting information on sex to collecting information on gender. This may or may not be accompanied by the collection of information on sex at birth in order to derive information about gender. While these concepts are often used interchangeably, they are distinct from sex.

900. The sex (male or female) of every individual should be recorded in the census or taken from the appropriate administrative record for countries using a register. Sex-disaggregated data is a fundamental requirement to understand variations in the experiences of males and females when it comes to a host of demographic, social and economic characteristics such as ageing, marital status, living arrangements, educational attainment, labour force status, migration and disability.

901. With growing awareness of the distinction between sex and gender, as well as the increasing visibility of gender minority populations, some countries have started collecting information on gender or gender identity in their census. Thus, at this time gender is recommended as a non-core topic for countries that want to measure the gender diversity of their population in a census. Because sex and gender are often conflated, it is important to make clear what is being measured.

902. Gender is broadly defined as a multidimensional concept that includes psychological, social and behavioural aspects. Gender identity is generally an individual’s self-perception or sense of belonging as a man/boy (masculine), woman/girl (feminine) or non-binary person (a person who is not exclusively man or woman). Transgender people are those who define their gender identity differently from their sex that was registered or recorded at birth.

903. There is diversity in the way in which countries approach inclusion of sex, gender or both sex and gender in data collection. This is highly dependent upon cultural context.

904. While the existing methods to collect information on gender demonstrate significant differences, all are supported by rigorous ongoing testing programmes. Countries have published robust research on topic development, question wording and response options based on knowledge from cognitive testing and focus groups. Information exists on field testing that includes examining collection methods by mode, with considerations for the privacy of respondents. Country collaborations have occurred to understand data processing methods and evaluations. Information outlining dissemination strategies is available that considers the delicate balance between confidentiality and the need for disaggregated gender data.

905. There are still many areas where additional gender research is needed. There is limited information about proxy reporting, where one person responds for all household members. Since research shows that gender identity forms at a variety of ages and that the number of people that one shares one’s identity with can vary, there are potential implications that are yet to be understood.

906. Another consideration is whether there are distributional differences when comparing responses for the core variable of sex compared to responses for either sex assigned at birth or gender. This consideration may become important where census data is used in downstream products such as intercensal estimates, population estimates or as survey controls. To measure a small population like the transgender population accurately, the alignment of sex and sex assigned at birth are relevant issues. Currently, Statistics Canada is using census data on gender as survey controls for their monthly Labour Force Survey.

907. Countries that have collected information on gender have used two different approaches.

908. Canada, Ecuador and New Zealand used a question design commonly referred to as the “two-step approach”, where a question on gender is paired with a question on sex at birth. Canada and New Zealand modernized their most recent censuses using gender as the default demographic for output. Canada now asks, “What was this person’s sex at birth?” as opposed to “What is this person’s sex?” as used in the 2016 census. This question’s response options of male and female remained unchanged across censuses. The next question asked, "What is this person’s gender?" with response options of “Male” and “Female” with a write-in option “Or please specify this person’s gender.”

909. Ecuador used a similar two-step approach, although they had more detailed response options for the gender question, including “Male”, “Female”, “Trans male”, “Trans female”, “Nonbinary” and “Does not know/ Does not respond.”

910. It is important to note that when using a two-step approach, it is necessary to use responses from both parts of the question to derive gender.

911. Censuses in the United Kingdom took a different approach, and this varied between countries. In England and Wales and in Scotland the core sex question was maintained, asking “What is your sex?”. Later in the England and Wales questionnaires, respondents were asked “Is the gender you identify with the same as your sex registered at birth?”, with response options of “Yes”, “No” and a write-in response. In Scotland, the second question asked, “Do you consider yourself to be trans or have a trans history?” with response options of “No” and “Yes, please describe your trans status”, with a write-in. The question includes the note “Trans is a term used to describe people whose gender is not the same as the sex they were registered at birth.”

912. The data from England and Wales showed that while there is confidence in the gender identity estimates at a national level, there are some patterns in the data that are consistent with – but do not conclusively demonstrate – that some respondents may not have interpreted the question as intended, for example, people with lower English language skills in some areas. At the time of preparing these Recommendations, findings have not yet been released in Scotland to determine how the second question asking about being “trans” or having a “trans history” performed.

913. In the New Zealand census, if the sex assigned at birth or gender responses did not include a usable value, information was gleaned from administrative data sources where the concepts align.

914. Countries using administrative data sources may need to consider the reference period for the given sex response. That is, does the recorded value represent one’s “sex at birth” or sex as of the reference period (i.e. end of year). Each country will need to weigh the importance of this distinction based on their national priorities and nationally available data.

915. While research on how to measure sex and gender is evolving, it is imperative that the existing validated research guide census data collections. This is particularly important for small and vulnerable population groups. Using the existing methods, it is important to determine whether collecting sex as a core variable is sufficient in the census or if information on gender is needed as a non-core topic. To inform this decision countries may consider whether collecting information on gender adds value within their policy, public opinion, and legal context given the sensitivity of the topic.

916. Information on date of birth should be collected to obtain information on age. This yields more precise information than collecting data on exact age at the census reference time. Collecting information on the date of birth allows for the tabulation of data in two ways: by year of birth and by completed years of age. Date of birth also allows age to be calculated relative to points in time other than the census reference time.

917. Since age is one of the most important topics collected in a census, used in many tabulations and analyses, information on age must be as complete and accurate as possible. If information on age is missing, imprecise or contradictory, imputation based on other individual or household entries should be undertaken.

918. In countries where the exact date of birth may be difficult to collect for a significant proportion of the population, the census questionnaire could allow calendar quarter or season of the year to be substituted. However, in census questionnaires with specifically designated response boxes for day/month/year, such write-in responses as season of the year would not be codable. In some cultures, newborn children may be considered to have age 1: in countries where this practice is common, care must be taken to distinguish these cases.

919. Impossible ages, such as children older than their parents, should be edited. If age in years is collected, attention must be given to parents mistakenly reporting their infant's age in months. Care should be taken in collecting and assuring the quality of information on the age of older persons, since misreporting can be more common at the older ages.

920. Children, young persons and older persons are age groups often requiring special focus in thematic analyses of census data. Any census characteristic in which age is expected to play a role (such as socioeconomic characteristics) should be readily cross-classified against age. Countries should ensure that the definitions and classifications used in the census meet the needs of likely users of data on these groups.

921. Legal marital status is defined as the conjugal status of each individual in relation to the marriage laws or customs of the country (sometimes also referred to as the de jure marital status).

922. Information on the legal marital status of each person should be collected at least for persons aged 15 years and over. However, since the minimum legal age (or the customary age) for marriage varies between countries, and since the population may also include young persons who have been married in other countries with lower minimum ages, it may be necessary to employ a lower age threshold for the collection of information on legal marital status.

923. The following basic classification of the population by marital status is recommended:

(1.0)Never married

(2.0)Married

(3.0)Widowed and not remarried

(4.0)Divorced and not remarried.

924. All persons living in consensual unions (see paragraph ‎934) should be classified as “never married”, “married”, “widowed” or “divorced” in accordance with their de jure (legal) status.

925. In countries with legal provision for registered/legal partnerships or marriage of same-sex or same-gender couples (depending on whether the legal provisions apply to sex or to gender), additional categories may be included in the category of the “married”, for example:

(2.1)Spouse of a different sex (or gender)

(2.2)Spouse of the same sex (or gender).

926. Alternatively, the classification may be expanded, for example, as:

(1.0)Never married

(2.0)Married

(2.1)Spouse of a different sex (or gender)

(2.2) Spouse of the same sex (or gender)

(3.0) Registered or other legal partnership

(3.1)Partner of a different sex (or gender)

(3.2) Partner of the same sex (or gender)

(4.0) Widowed and not remarried

(5.0) Divorced and not remarried.

927. In some countries there may be a policy or user need to distinguish male same-sex couples and female same-sex couples. This can be achieved by considering the combination of sex and legal marital status according to this expanded classification.

928. The term “spouse” may be replaced, in collection instruments and/or in dissemination products, by terms appropriate to the language and the culture of each country, such as “husband/wife”. Many languages lack a gender-neutral word for “spouse”, or the word may carry certain connotations that may affect understanding or willingness to answer. Thorough testing for acceptability and impacts on data quality should be conducted, both when introducing any expanded classification to account for same-sex or same-gender partnerships, and when choosing terms to be used in questionnaires and for dissemination.

929. A separate category for “legally separated” may be necessary in countries where the legislation includes provisions for this status, as distinct from “married” or “divorced”. Some countries may also wish to identify separately within the “married” category, those persons who are remarried.

930. Unless a separate category is used for annulled marriages, individuals whose marriages have been annulled should be classified according to their marital status prior to the annulled marriage.

931. De facto marital status is defined as the status of each individual in terms of their actual living arrangements within the household being enumerated, regardless of that person’s legal marital status.

932. Information on this topic should be collected for persons of the same age categories as those for whom information on the legal marital status is collected.

933. The recommended classification is:

(1.0)Person living with spouse or partner (registered or in a consensual union) having usual residence in the same household

(2.0)Person not living with spouse or partner (registered or in a consensual union) having usual residence in the same household.

934. Two persons are taken to be partners in a consensual union when: they have usual residence in the same household, they are not married to, nor in a registered/legal partnership with, each other, and they have a marriage-like relationship to each other. People in such a relationship are also known as “cohabiting partners” or a “cohabiting couple”.

935. Some countries may wish to facilitate analysis of non-cohabiting marriages and consensual unions; that is, where the criteria above are fulfilled with the exception of having usual residence in the same household (or where a couple is legally married but not having usual residence in the same household); in which case countries should add to the above classification (3.0) “Person with spouse or partner (registered or in a consensual union) not having usual residence in the same household”.

936. An optional distinction may be made within category (1.0) between (1.1) “Person living with spouse or partner of a different sex (or gender)” and (1.2) “Person living with spouse or partner of the same sex (or gender)”.

937. Information on de facto marital status can also be derived from information collected on topics related to household and family characteristics of persons, characteristics of family nuclei and characteristics of private households, based on the relationship to the reference person or the full household relationship matrix where used (see section ‎16.3.2.2). Where such a matrix is not used, it is necessary in censuses employing direct enumeration to ask a specific question on living arrangements.

938. In countries where census information is derived from administrative registers, data on consensual unions and non-cohabiting unions may not readily be available.

939. Information on the total number of children born alive can be collected in the census in countries in which such information is needed for estimating fertility using indirect techniques. In countries with unreliable or incomplete registration of births, census information on this topic can be useful for assessing the completeness of the registers and for estimating levels of lifetime fertility of older cohorts.

940. This topic can be sensitive for mothers who have experienced stillbirths, as well as for those who have given birth outside of a socially sanctioned union. This should be considered carefully in the testing programme. If included in the census, information on total number of live-born children should be collected for all women aged 15 years and older. In populations where significant fertility occurs at younger ages, the lower limit should be adjusted accordingly.

941. The data collected on total number and sex of live-born children should, in principle, include all children born alive during the lifetime of the women concerned up to the census date (that is, excluding foetal deaths and stillbirths). Adopted children should not be included. The number recorded should comprise all live-born children, whether born within the present marriage or other form of partnership or a previous one, whether born within a union or outside of one, within the country or abroad, and regardless of whether such children are still alive at the census reference time or where they are living.

942. Information on dates/duration of marriage can be valuable for fertility statistics, and extends the knowledge that can be derived from data on number of live-born children. In the case of women who have been married more than once, information may be obtained on the dates of both the first marriage and the current marriage.

943. For women who are widowed, separated or divorced at the census reference time, “date of/age at/number of years since the dissolution of first marriage” should be collected. Information on dissolution of first marriage (if applicable) provides the data necessary to calculate “duration of first marriage” as a derived topic at the processing stage. In countries in which duration of marriage is reported more reliably than age, tabulations of children ever born by duration of marriage yield better fertility estimates than those based on children born alive classified by age of the woman. The duration of marriage can be obtained by subtracting the age at marriage from the current age, or directly from the number of years elapsed since the marriage took place. Some countries may find it useful to also collect the order of the current marriage.

944. Information on dates/duration of consensual union(s) can be valuable for the same reasons as the corresponding information on marriage(s) (paragraph ‎942).

945. In the case of women who have been in consensual union more than once, it is recommended to obtain information on the dates of both the first and the current consensual union. Information on dates/duration of consensual union(s) can be combined with the information on the date(s) of legal marriage(s).

946. It should be recognized that the beginning, and thus the beginning date, of a consensual union may be imprecise.