Data on families, cohabitants and spouses

The data collection shows the status of family, spouse and cohabitant relations for residents of Norway, i.e. those who have a personal identification number, for the period 1975 to the most recent available year. There is one entry for each person, and their status is updated for each year that they have been resident according to the CPR. Data is organised in a time series file that is updated annually, and people who have died or have emigrated remain included in the data.

The file shows family and spouse relations from 1975 and cohabitant relations from 1987 up to the most recent available year. The family number is based on cohabiting couples as well as married couples living together, from 2005.

The file does not have official statistics status. It has some gaps and shortcomings, which are detailed below. When the file is used, new irregularities may sometimes come to light. Statistics Norway would like users of the file to let us know if they come across any new ambiguities. Despite this problem, the file contains a lot of valuable information, which is undoubtedly of great benefit to those who need to follow the relations within the population over time.

The data collection contains relations between people, using personal identifiers. When data is made available for research, all variables that can identify individuals are replaced with a serial number. The serial number allows users to follow the individuals and their relations over time.

1. Content of data collection

The statistical unit in the data collection is the statistical ID number (a statistically non-meaningful ID created in Statistics Norway), and the population comprises of all valid statistical ID numbers that also have a personal identification number, as per the production date. The personal identification number is assigned at birth or upon settlement in Norway, i.e. the population does not include persons who are not currently or have not previously been resident (those with no D number).

For each year, the file is updated with a new year for the statistical ID number, personal identification number, family number, marital status, cohabitant’s statistical ID number, spouse’s personal identification number and spouse’s statistical ID number. The variables for the new year are left blank for persons who are no longer resident in Norway (deceased or emigrated).

2. The variables

Statistical ID number:

As already mentioned, the statistical ID number is a statistically non-meaningful ID that is used internally in Statistics Norway.

The statistical ID number is a serial number that uniquely identifies people over time. When a personal identification number is changed one or more times, the statistical ID number will be the link between the different personal identification numbers and will ensure that we have a unique key over time.

The statistical ID number should, in principle, have a one-to-one relationship to individuals. However, there have been a few cases where two people have had the same statistical ID number. In such instances, the person who was first assigned the statistical ID number gets to keep it, while the other gets a new one. This applies to around 300 of the 7.4 million people who were included in the file as of 1 January 2016. The errors this causes, both in relation to the individual concerned and their cohabitant/spouse, are not corrected in the file.

Cohabitant’s statistical ID number

Where a person lives with their spouse, this variable is the same as ‘spouse’s statistical ID number’. The variable is blank if the person does not have a cohabitant or does not live with a spouse, and returns to blank upon the break-up of a cohabitant relationship.

Cohabitant relationships for the period 1987–2004 are entered in the statistics based on status in 2006, which is the year after cohabitants with no children together were first considered to be cohabitants in the statistics. In the file, the current cohabitant definition is used for the entire period from 1987 to the present.

Family number

The family number used in family statistics is the personal identification number of the oldest person in the family. Family is currently defined as persons residing in the same dwelling who are connected to each other as spouses, registered partners, cohabitants and/or as parents and children (regardless of the age of the children). A family can consist of a maximum of two generations. If a child living in the family home has a child, this child and its parent will constitute a new family with their own family number. Everyone has a family number, including people who live on their own. A household may consist of several families (who live in the same dwelling).

• In the period 1975–1991, only spouses and their children were classed as the same (statistical) family and had the same family number.
• In the period 1992–2004, cohabitants with children together were also classed as the same family in the statistics.
• As from 2005, cohabitants without children together were also connected to each other with the same family number.

In this data collection, Statistics Norway’s family formation from 2005 is used to define cohabitants in data dating back to 1987. Those in the file are grouped together in cohabiting couples using current methods dating back to 1987.

However, the family number is not up-to-date in the file, which contains the family number from the CPR up to the end of 2004. In the CPR family number, neither cohabitants with or without a child together are included (in contrast to Statistics Norway’s system). The identification of cohabitants with children together from 1992 up to the end of 2004 is thus not used to give these cohabitants a common family number. This first takes effect in the file in 2005. Since 2005, cohabitants (with and without children together) have had a common family number.

Marital status

Marital status as of 1 January each year. From the CPR. Obtained from the CPR. The code values for marital status are available at ssb.no.

Marital status is not linked to cohabitation (the variable ‘Cohabitant’s statistical ID number’). Two people may be cohabitants (also in the statistics) even in cases where both are married to someone else.

Spouse’s personal identification number

The value remains blank until a marriage is entered into. From 1975 up to the end of 1985, the spouse’s personal identification number and statistical ID number remained in place after divorce or the death of a spouse until the person remarried or entered into a new partnership. In the period 1986–1995, the value went back to being blank after divorce or if the spouse died. In 1996, the pre-1986 practice was reintroduced (personal identification number and statistical ID number remain).

For changes to the personal identification number, this only applies to the person in question. If the person is someone’s spouse, the ‘Spouse’s personal identification number’ for this person is not updated for earlier years. Every year, 30–70 people receive a new personal identification number.

There will always be some spouses with an invalid personal identification number. These consist of just the date of birth and do not include the personal number part of the personal identification number.

Spouse’s statistical ID number

Spouses with an invalid personal identification number are not given a statistical ID number. Therefore, there will always be a few thousand spouses with an invalid personal identification number who do not have a statistical ID number. Figure 1 reflects this situation, as well as the transition from 1995 to 1996 as described above, where divorcees and widows/widowers went from having a blank code to keeping the personal identification number/statistical ID number until they remarried or started cohabiting. Note that the scale in the figure does not start at zero, but 1 500 000.

Figure 1. Spouse’s personal identification number and statistical ID number

3. Quality

Uncertainty in the CPR

The CPR is the basis for all data in the file. There will never be full concordance between the address registered in the CPR and the actual address for 100 per cent of the population. First, there is the situation where students who do not live in the family home can report their relocation if they want, but are under no obligation to do so. Therefore, many students are still shown as living with their parents in the CPR.

Second, not everyone reports their relocation to the CPR, including internal migrants, and perhaps especially immigrants who emigrate again, who often do not see any reason to inform the Norwegian authorities of their move. Even though the Directorate of Taxes reviews the register periodically, there will always be a certain amount of people who are either registered at the wrong address within the country or who are registered as resident in Norway even if they no longer live here.

A third group is made up of people who are living legally in the country on short-term stays, and who have a D number but not a personal identification number. Those living in Norway illegally constitute the fourth group.

Imputing couples and families

Statistics Norway’s primary task is to produce statistics. Grouping people into families and households is reasonably unproblematic, but uncertainties can occur. Couples who are married to each other (and have children together) and are registered at the same address are easy to place. Cohabitants (at the same address) with children together are also unproblematic. Cohabitants without children together are slightly more difficult to place. When these are to be grouped in families, this is done using statistical methods. We assume that if certain conditions exist, two people are connected to each other in a couple relationship, which means that they belong to the same family in our statistics. At an individual level, this is bound to be wrong in some cases. However, as long as the purpose is to produce statistics, the errors will balance each other out and the uncertainty is unproblematic.

However, when data is to be used at an individual level, the uncertainty in the data needs to be taken into account. Some people who are shown as cohabitants in the file are not actually cohabiting.

Old errors in couple formations

The last major change in the family and household statistics took place in 2005. This change saw cohabitants without children together being included in the statistics as cohabitants for the first time. Based on the situation as of 1 January 2006, the new family/cohabitant formation was applied to those in the cohabitant, family and spouse files dating back in time. Thus, it is the current definition of cohabitant and family that provides the basis for data on cohabitants and families all the way back to 1987. However, the work in 2006 was subject to errors. Figure 2 shows the number of cohabitants who were not married to each other in the file for the entire period 1987–2014.

Figure 2. Number of cohabitants

Source: Statistics Norway

Between 1987 and 2000, there was a steady increase from year to year, from 136 000 cohabitants (not couples) in 1987 to 361 000 in 2000. The number then dropped to 303 000 in 2001 and 277 000 in 2002, before increasing again by 100 000 to 377 000 in 2003. We have every reason to believe that this is not correct.

Figure 3. New cohabitants

Source: Statistics Norway

Figure 3 shows new cohabitants for each year, i.e. people who were cohabitants and were not married to the person they were living with at the end of the year, and were not cohabiting with the same person at the start of the year. This encompasses persons who did not have a cohabitant the year before as well as persons who were cohabiting with someone else the year before. This figure is also doubtful. From a level of approximately 60 000 new cohabitants annually up to 1999, there was a sudden dip to 11 000 in 2000 and 8 000 in 2001 before the gap was closed again, with 123 000 new cohabitants in 2002. There were also major fluctuations in 2003 and 2004.

The cohabitants are ‘coupled’ correctly, but in the period 2000–2004, many people’s new cohabitation relations were registered at the wrong time.

Overall, the file provides credible information about families, cohabitants and spouses. However, the data quality is not sufficient to capture when changes in cohabitation relations took place in the period 2000–2004.

Inconsistencies in the cohabitant variable between years

One observant user pointed out that there was a remarkably high number of people who had the same cohabitant in 2009 and 2011, but no cohabitee in 2010. There is a normal variation of establishment, termination and reestablishment of cohabitation, with a population of approximately 3 000 people per year. However, for 2010, this figure was much higher. This turned out to be incorrect and was corrected in the current version of the file (dated 19 November 2016).