Comparability — Botswana [top]

In 1981, a systematic sample was selected from a 25% random sample, so the method identified is "systematic".

Comparability — Canada [top]

The 1971, 1981, and 1991 samples have a systematic selection and geographic stratification.

Comparability — Colombia [top]

The 1985 sample is a systematic sample of dwellings, where the selection was performed separately for urban and rural areas.

The 2005 sample followed a systematic selection of every other dwelling from the original 20% microdata. The original 2005 20% sample is a Bernoulli stratified design for elements of households, Poisson for dwellings, and of conglomerates for persons. The resulting 2005 sample is classified as stratified systematic in SAMPMETH.

Comparability — Egypt [top]

Comparability — Ethiopia [top]

The sampling method for the 1994 sample is a 1-in-2 systematic sample drawn from the original 20% systematic sample for all regions except Somali and Affar, which were sampled at lower densities.

Comparability — France [top]

The 1990 and 1999 samples are both systematic samples with stratification based on locality and household size; the 1990 data also consider district clusters.

The 2006 and 2011 samples are stratified clustered with one-fifth of communes under 10,000 population taken in their entirety and 8% of housing units sampled from 10,000 or more population communes. The microdata are a 25% sample for smaller communes and a 40% sample in communes with population over 10,000.

Comparability — Germany [top]

The 1970 data is a random sample drawn from the 10% scientific use file (SUF), thus it is classified as a simple random sample. The 1971, 1981, and 1987 samples used systematic selection on geographically sorted data, by including households for which the last two digits of the household serial number was part of a systematically selected set of numbers 0-99.

Comparability — Haiti [top]

The 1971 sample is systematically drawn for the urban population; while the sample for the rural population is 10%, but the sampling method is not specified. In 2003, the sampling method is systematic, selecting every 10th household with a random start.

Comparability — Honduras [top]

The sampling method for 1988, 2001, and 2013 samples are all systematic samples, selecting every 10th household with a random start in the case of 1988, and selecting every 10th dwelling with a random start in the case of 2001 and 2013.

Comparability — Indonesia [top]

The 1971-2000 samples are stratified. The stratification criteria correspond to population size, population density, or urban/rural status in 1971-1995 and 2005. The 2000 sample is geographically stratified by virtue of households being sorted by geographic location prior to drawing a systematic sample. The 2010 sample was drawn by the national statistical institute, but the sampling method is unknown.

The 1971-1995 and 2005 samples are clustered, covering selected census blocks or enumeration areas.

Comparability — Ireland [top]

The 1971-2016 samples are a 10% random sample of the household records from each county, making counties the implicit geographic strata for each sample.

Comparability — Israel [top]

In 2008, the sample that received the questionnaire is composed of subdivisions of statistical areas (containing approximately 50 households) that were randomly selected. Thus, the sample is classified as a cluster sample.

Comparability — Jamaica [top]

The 2001 long form was applied to a selection of 10% of all enumeration districts and it was classified as "cluster".

Comparability — Jordan [top]

The 2004 sample is stratified considering large cities, urban, and rural areas, where the selection of households was done systematically.

Comparability — Mauritius [top]

The 1990, 2000, and 2011 samples are a systematic sample of every 10th household after a random start.

Comparability — Mongolia [top]

The 1989 sample is a cluster sample that only includes some districts (soums), whereas the 2000-2020 samples are systematic samples of households with full geographic coverage.

Comparability — Pakistan [top]

The 1981 sample is a systematic sample of person records, drawn from a 38% sample of short-form data. The 1981 sample was drawn systematically by IPUMS from an original 38% sample drawn by the national statistical office, but no information is available about the design of the original sample, so the sampling method is coded as "Unknown". The 1998 sample excludes three provinces: Fata, Northern Areas, and Kashmir. In 1981 and 1998, persons are not organized into households.

Comparability — Palestine [top]

The national statistical office constructed 20% systematic samples of households from the 1997, 2007, and 2017 censuses; then, 1-in-2 households were selected systematically from each original sample.

Comparability — Panama [top]

In 1970, the data is a 1-in-2 systematic selection drawn from the sample created by the national statistical institute, but the sampling method for the original sample is unknown.

Comparability — Peru [top]

The 2007 sample was systematically drawn from the 15% stratified sample developed by the national statistical institute.

Comparability — Puerto Rico [top]

The 1970 census corresponds to data from the long form, which was assigned to respondents systematically; then, a sample was drawn systematically from the long form data using stratification based on population characteristics. The 1980, 1990, and 2000 census long form samples had a sampling rate depending on the population size of the area to be enumerated, where households were to be selected systematically; similarly, the public use samples created from them followed a stratified systematic selection procedure.

The Puerto Rico Community Survey 2005, 2010, 2015, and 2020 had a stratified sampling design with systematic selection of households.

Comparability — Rwanda [top]

The 1991 and 2002 samples are a systematic selection of every 10th dwelling with a random start. The 2012 method is based on an equal probability of selection for all private households, but the specific sampling method is not specified.

Comparability — Senegal [top]

A systematic sample of every 10th household with a random start was drawn for the 1988 and 2002 samples.

Comparability — Sierra Leone [top]

The 2004 and 2015 samples are systematic samples of every 10th household with a random start.

Comparability — South Sudan [top]

In the 2008 census, the long form was applied to systematically selected enumeration areas within each state, stratified by urban and rural area of residence.

Comparability — Spain [top]

The 2011 census is classified as "systematic", but the sample design also contemplated that some dwellings were to be selected through Bernoulli sampling.

Comparability — Sudan [top]

In the 2008 census, the long form was applied to systematically selected enumeration areas within each state, stratified by urban and rural area of residence.

Comparability — Tanzania [top]

The 1988 and 2002 samples are a systematic selection drawn from the original long form data; the long form data followed a stratified systematic design within sampled enumeration areas.

Comparability — Thailand [top]

The 1970, 1980, 1990, and 2000 census samples are stratified by region and municipal and non-municipal areas. The 1990 and 2000 samples were drawn in two stages, with enumeration districts sampled in the first stage, and households within selected enumeration districts in the second stage. The 2000 sample documentation further specifies that households within enumeration areas were selected systematically.

Comparability — Togo [top]

The sampling method for the 1960 sample is a systematic sample of every 10th household in seven urban areas.

Comparability — Uganda [top]

In 1991, the long form was applied to a sample stratified by urban and rural location, which included all urban enumeration areas and approximately 10% of rural enumeration areas; a sample of households was drawn systematically from the original data for urban areas, to obtain an overall 10% sample for the country.

Comparability — United Kingdom [top]

The 1961 and 1971 data follow a similar sample design: the original data is a 10% sample that was then stratified by household type, and a random sample extracted from it to obtain 1% of the population. The 1961 original 10% data correspond to the long form, which was systematically selected during the data collection process.

The 1991 and 2001 samples are stratified by enumeration district.

Comparability — United States [top]

The 1960 and 1970 censuses correspond to data from the long form, which was assigned to respondents systematically; then, samples were drawn systematically from the long form data using stratification based on population characteristics. The 1980, 1990, and 2000 census long form samples had a sampling rate depending on the population size of the area to be enumerated, where households were to be selected systematically; similarly, the public use samples created from them followed a stratified systematic selection procedure.

The American Community Survey 2005, 2010, 2015, and 2020 had a stratified sampling design with systematic selection of households.

Comparability — Uruguay [top]

The 1963, 1975, 1985, 1996, 2007, and 2011 samples are systematic samples of every 10th household with a random start.

The 2006 data correspond to the Extended National Survey of Households (ENHA). The sample is stratified, with systematic selection of clusters and households.