The I4NG Harmonised Data: Demographics! - Documentation
This document contains all relevant information on the demographic variables included in the I4NG Harmonised Data: Demographics!.
For each variable, the target categories are listed, followed by a short note on decisions made where the source surveys required a harmonisation choice. These notes should be taken into account when utilizing or interpreting the demographic variables.
Missing Values
The following represents the general logic and common usage of the missing value codes across the harmonised demographic variables. Additional specificities are explained in the respective variable sections.
-9 “NAs from source data sets” covers source codes, such as “No answer,” “Don’t know,” or “Refused”, cases where the respondent did not, or could not, provide a substantive answer.
-999 “Inap. from source data sets” is the code for all inapplicable data in the source data set. This covers cases where the underlying question was not asked in a given country, where the source survey never fielded a comparable question at all, and where the source data are blank with no user-defined value label (system-missing).
bv_sex
Variable label: Sex.
| Code | Label |
|---|---|
| 1 | Male |
| 2 | Female |
Variable name in the harmonised dataset: bv_sex.
This variable does not include a third, non-binary category. In the very rare cases where a source survey offers such a category (e.g., recent EB waves), it was recoded to -9
Cases coded to -999:
- ISSP 2001 includes blank cells in the source data with no user-defined value label.
bv_age
Variable label: Age.
| Code | Label |
|---|---|
| Integer | Age in years |
Variable name in the harmonised dataset: bv_age.
In earlier ISSP waves such as 1985, 1987, 1992, Italy and Denmark provided age categories instead of exact age values. These were recoded to the midpoint of each category.
Cases coded to -999:
- A small number of dubious age values identified in EB wave 93.2.
- Blank cells with no user-defined value label in ISSP 1997.
Sampling universe regarding age: Across the survey programmes, the target populations differ in their age eligibility criteria. Eurobarometer (EB) and the European Social Survey (ESS) generally target populations aged 15 and over, while the European Quality of Life Survey (EQLS), European Values Study (EVS), and International Social Survey Programme (ISSP) generally target populations aged 18 and over. These general age criteria are subject to certain deviations across individual waves and survey programmes.
The harmonised data contain differences in the observed age ranges across survey programmes.
Age distributions by survey programme are shown below.
| Survey | N | Mean age | Min. age | Max. age |
|---|---|---|---|---|
| Eurobarometer (EB) | 2,713,236 | 48.5 | 15 | 99 |
| European Quality of Life Survey (EQLS) | 142,435 | 49.3 | 18 | 95 |
| European Social Survey (ESS) | 488,274 | 48.6 | 13 | 123 |
| European Values Study (EVS) | 216,369 | 46.3 | 15 | 108 |
| International Social Survey Programme (ISSP) | 1,388,062 | 46.9 | 14 | 112 |
bv_age_cat1
Variable label: Age by category: “Young Europeans” < 35.
| Code | Label |
|---|---|
| 1 | Under 35 |
| 2 | 35 and older |
Variable name in the harmonised dataset: bv_age_cat1.
- The purpose of creating age categories is to facilitate the analysis in line with the I4NG data summaries. In this variable, the first category includes all respondents younger than 35. This definition of “Young Europeans” corresponds to the definition used in the I4NG Data Summaries; please see here.
bv_age_cat2
Variable label: Age by category: “Young Europeans” < 25.
| Code | Label |
|---|---|
| 1 | Under 25 |
| 2 | 25-34 |
| 3 | 35-54 |
| 4 | 55 and older |
Variable name in the harmonised dataset: bv_age_cat2.
- The purpose of creating age categories is to facilitate the analysis in line with the I4NG data summaries. This variable distinguishes four age groups: below 25, 25–34, 35–54, and 55 and older.
bv_education
Variable label: Education - Low/Middle/High derived from the highest completed degree.
| Code | Label |
|---|---|
| 1 | Low |
| 2 | Mid |
| 3 | High |
Variable name in the harmonised dataset: bv_education.
Target categories were derived using ISCED levels as the basis where available. “Low” education covers degree up to the lower secondary level, “Middle” covers all higher secondary degrees below the university level and the category “High” covers the tertiary education sector.
For the surveys where no information on respondents’ degrees is available, the variable bv_educyears was provided. It offers information on the age when education was completed.
ESS: uses the pre-harmonised ES-ISCED variable. The residual “Other” category was recoded to -999.
EVS: The source variable differs by wave. The 2017 wave provides ISCED codes that are collapsed to the three target categories. The 2008 and 1999 waves provide a pre-recoded three-category variable used directly.
The cases of the following datasets were coded to -999 because:
- EB does not provide a variable for the highest educational level attained.
- No comparable education variable exists for EVS 1990 and 1981.
- ESS 2002–2008 waves contains code 0, “not possible to harmonise”.
- Source categories that combine “still at school” with other non-degree statuses (e.g. ISSP 2001: “None; still at school, uni”), since these cannot be assigned to a target category.
- Blank cells with no user-defined value label occur in ISSP 2001.
bv_educyears
Variable label: Education information for EB and EVS (1981, 1990): Age when school completed.
| Code | Label |
|---|---|
| -1 | Still studying (missing for this variable) |
| 0 | Never attended school, or no schooling |
| 12 | 12 years or less |
| Integer | Age in years 13-20 |
| 21 | 21 years or more |
Variable name in the harmonised dataset: bv_educyears.
Values are between 12 and 21, 12 = 12 years or less, and 21 = 21 years or more.
The cases of the following datasets were coded to -999 because:
- EB waves 74.1, 73.2, 56.2, 43.1, 29, 25, and 18 start at “16 or less”. They were not harmonisable.
- Cases of all datasets except EB and EVS (1981, 1990) that has bv_education.
bv_status
Variable label: Main (employment) status.
| Code | Label |
|---|---|
| 1 | Paid work |
| 2 | In education |
| 3 | Unemployed |
| 4 | Permanently sick/Disabled |
| 5 | Retired |
| 6 | Community or military service |
| 7 | Housework, looking after children or others |
Variable name in the harmonised dataset: bv_status.
EB does not have the source categories for the target codes 4 and 6, they are entirely absent from the EB data.
EB does not distinguish “permanently sick/disabled” from “retired” such that both fold into “Retired”.
ESS: Source variable which split unemployment into “looking” and “not looking” for a job were merged into code 3.
EVS 2017 and 2008: three paid-work categories (full time, part time, self-employed) all were mapped to code 1.
EVS 1999, 1990, 1981: disabled and military service categories do not exist.
A residual “Other” category, where present in ESS and EVS, was recoded to -999, as its content cannot be reliably attributed to a target category.
The cases of the following datasets were coded to -999 because:
- EVS 1990 (Malta) did not ask a question about the employment status.
- EB waves 29, 25, and 18, where students and military service fall into the same source category and cannot be distinguished.
- EB wave 57.2 has no data for Poland.
- EB wave 59 uses a 3-category scheme where “not working” cannot be split into Unemployed/Retired/Housework/Student.
- ISSP 1997 and 2001 include blank cells with no user-defined value label.
bv_urbanrural
Variable label: Place of living urban - rural.
| Code | Label |
|---|---|
| 1 | Urban |
| 2 | Suburban |
| 3 | Rural |
Variable name in the harmonised dataset: bv_urbanrural.
The cases of the following datasets were coded to -999 because:
- EVS 1981 Denmark did not ask a question about the place of living.
- EVS 1990–2017, where no comparable urban/rural variable exists for any of the four waves.
- EB waves 54.2 and 51.1 have no urban/rural item at all.
- ISSP 1992, 1997, 1999, and 2001 include blank cells in the source data with no user-defined value label.
bv_marital
Variable label: Legal partnership status.
| Code | Label |
|---|---|
| 1 | Married |
| 2 | Registered partnership |
| 3 | Separated/Divorced |
| 4 | Widowed |
| 5 | Never married/Never registered partnership |
Variable name in the harmonised dataset: bv_marital.
This variable reflects legal partnership status only. Cohabitation or “living as married” was not treated as equivalent to a registered partnership and was recoded to code 5 (Never married/Never registered partnership).
Separated and Divorced were combined into a single category (code 3, “Separated/Divorced”).
The cases of the following datasets were coded to -999 because:
- ESS 2008 and 2010 did not administer this item in Estonia and Finland, respectively.
- EVS 1990 and 1981, where no comparable partnership variable exists.
- EB waves 57.2 (no data for Poland) and 54.2 (item fielded only in Norway) carry a country-specific “Inap.” code.
- ISSP 1997 and 2001 include blank cells in the source data with no user-defined value label.
bv_denomination
Variable label: Religious affiliation.
| Code | Label |
|---|---|
| 0 | No religion |
| 1 | Catholic |
| 2 | Protestant |
| 3 | Orthodox |
| 4 | Other Christian |
| 5 | Jewish |
| 6 | Islamic |
| 7 | Eastern/Asian religions |
| 8 | Other Religions |
Variable name in the harmonised dataset: bv_denomination.
Hindu and Buddhist categories, whenever available, were collapsed into code 7 (Eastern/Asian religions).
“Not applicable” filters, indicating that the respondent does not belong to any denomination, were coded to 0.
“No denomination given” were recoded to -9 ISSP 2001 and earlier years.
The cases of the following datasets were coded to -999 because:
- The source data has blank cells with no user-defined value label (ESS 2002, 2004, and 2006).
- EB waves 25 and 18, where the source “Other” category combines several denominations of which some are and some are not part of the target coding frame (e.g. Jewish, Other Christian, Eastern religions) and therefore cannot be mapped to the code 8 (Other Religions).
- ISSP 1997 and 2001 include blank cells in the source data with no user-defined value label.