Comparability — Argentina [top]

The 1991-2001 classifications are fundamentally different from the earlier samples and from other countries, and they could not be incorporated into OCCISCO.

Comparability — Armenia [top]

The sample uses the ISCO 1988 classification, requiring no recoding.

Comparability — Austria [top]

The 1991 and 2001 samples used the ISCO classification. The earlier samples were recoded based on occupational title. The change of classifications between 1981 and 1991 is associated with a reassignment of cases from professionals to technicians.

The source data for Austria assigned occupations to non-workers based on who "supported" them. OCCSICO reports data only for economically active persons.

Comparability — Belarus [top]

The original data for Belarus used the ISCO system, requiring no recoding.

Comparability — Benin [top]

The 2013 sample uses the ISCO 1988 classification, while the earlier samples use ISCO 1968. The universe varies across samples.

Comparability — Bolivia [top]

The underlying data are consistent across the 1992-2012 samples, but the 1976 classification was notably different. The 2012 sample uses the ISCO 2008 classification system. The age universe is consistent across samples, but the 1976 and 1992 samples include some or all unemployed persons in the universe, while the 2001 and 2012 sample do not.

Comparability — Botswana [top]

The 1981 sample follows the ISCO 1968 classification system. In 1981, it was not possible to distinguish technicians from professionals due to the level of disaggregation of occupations available for the sample. The 1991-2011 samples use a classification system that is the national version of ISCO 1988.

The reference period for the 1981-1991 samples was the past month, but it changed for the 2001-2011 samples to the week prior to the census.

Comparability — Brazil [top]

The age universe is consistent across samples, but employment status and reference period for individuals included in the universe changes over time.

The 2010 sample followed the ISCO 2008 coding scheme. Earlier samples did not use the ISCO classification system and required more extensive recoding.

Comparability — Burkina Faso [top]

The 1985 and 1996 samples used the same non-ISCO coding scheme. The codes were grouped hierarchically by industry and level of seniority. See the unharmonized source variables for full detail.

Comparability — Cambodia [top]

The original data for all Cambodia samples were coded into the ISCO 1988 or 2008 systems, requiring no interpretation at the one-digit level. They report main activity last year.

Comparability — Cameroon [top]

The 1976 sample used a non-ISCO coding system, while the 2005 sample used the ISCO 1988 system.

The 1976 classification scheme is organized by industry and skill level, but it is difficult to make some distinctions required by the ISCO scheme.

Comparability — Canada [top]

The Canada sample data do not have much occupational detail (except 2011) and are based on national occupational classifications not fully equivalent to ISCO, so the coding to OCCISCO is particularly crude.

The classifications for 1971 to 1991 were quite consistent. Technicians are not identifiable in 1971-1991.

Comparability — Chile [top]

The age universe changes over time. There is a large increase in the technician category between 1992 and 2002.

Comparability — China [top]

The Chinese occupational classification was rather industrial in character: it often reported the good or service in which a person was engaged in producing without giving that person's status or precise job. It was particularly hard to identify persons who unequivocally belonged in the elementary (unskilled) occupations (code 090). Skilled agricultural and craft workers are probably inflated, and elementary workers underrepresented, in the OCCISCO classification.

The China 2000 sample has a more aggregated occupational grouping available with respect to 1982 and 1990, which produces additional comparability issues. In 2000, categories corresponding to professionals may include some of the related technical occupations. The proportion of persons in elementary occupations in 2000 is considerably smaller than previous samples; some of these cases are seemingly included in "service workers and show and market sales" or "crafts and related trades workers."

Comparability — Colombia [top]

The basic classifications of occupations for Colombia 1964 and 1973 differ, but not substantially. There is no occupation variable for 1985 and 1993.

Colombia 1973 has many undocumented values. In most cases we coded them into occupational groups on the basis of their first digit(s).

Comparability — Costa Rica [top]

The variable records the person's main occupation at the time of census or, prior to 2000, when they were last employed. OCCISCO is not available for 1963 because we lack the necessary documentation.

Comparability — Cuba [top]

The 2012 sample uses the ISCO 2008 classification system, and the 2002 sample uses a national classification that is relatively consistent with the ISCO coding structure. The universe is consistent across the two samples, and the occupation categories are generally comparable.

Comparability — Côte d'Ivoire [top]

The occupation labels in the 1998 sample were originally in French and based on the ISCO88 classification.

Comparability — Dominican Republic [top]

The 1960, 1970 and 1981 samples each use a unique classification system. The 2002 and 2010 samples use the ISCO classification system.

The universe changes over time. In 1960, the question was asked of persons age 10 and older who were employed between January 1st and August 7th, 1960. In 1970, the question was asked of persons age 10 and older who were employed or unemployed the week prior to the census. In 1981-2010, the question was asked of persons age 10 and older who had ever worked. Respondents who had worked before but were not working during the week prior to the census reported their last occupation.

Comparability — Ecuador [top]

The age universe changes over time. The underlying codes from which OCCISCO was derived differ for each sample.

Comparability — Egypt [top]

The 1986 sample uses the ISCO 1968 classification, while the 1996 and 2006 samples use ISCO 1988. Thus, the data required no recoding, but the age universe of respondents changes from 15 and older in 1986-1996 to 6 and older in 2006.

Comparability — El Salvador [top]

Both samples use the ISCO classification and differ only in coverage of unemployed persons.

Comparability — Ethiopia [top]

The 1984 sample uses the ISCO 1968 classification, requiring some recoding. The 1994 sample uses ISCO 1988. The 2007 sample does not contain occupation information.

Comparability — Fiji [top]

The 1976 and 1986 samples both used the ISCO 1968 classification system. The 1996 and 2007 samples categorized occupation according the ISCO 1988 system, thus no recoding was necessary. The 2014 sample used ISCO 2008, requiring limited interpretation.

Comparability — Finland [top]

The 2010 samples uses the ISCO 2008 classification system.

Comparability — France [top]

Occupations for all French samples were integrated by the French statistical office into a common classification for the years between 1962 and 1990. Because the French occupation variable is not very detailed, difficulties sometimes arose in making the data correspond with the ISCO system. During those years, the elementary workers category for France contains only agricultural workers; persons with non-agricultural occupations that elsewhere might be classified as elementary workers cannot be distinguished. The armed forces category includes police.

In 1999, the French occupations accord with the ISCO coding scheme. However, in 2006-2011 the data again change to a classification that corresponds poorly with ISCO in a number of areas. It should be considered a very rough approximation; see the unharmonized source variables for the original codes.

There is also slight variation in the age universe over time.

Comparability — Germany [top]

The samples use different classifications.

The 1981 sample (from East Germany) is highly industrial in character and often does not distinguish skill levels in a manner that is well suited to the ISCO system. It should be considered a very rough approximation; see the unharmonized source variable for the original codes.

Comparability — Ghana [top]

The 1984 and 2000 samples use the ISCO 1968 classification system. The 2010 sample uses the ISCO 1988 classification system. The universe differs slightly across samples in terms of minimum ages.

Comparability — Greece [top]

The same occupational code system was used for the 1971-1991 censuses. The 2001 census used a version of the ISCO 1988 classification, while the 2011 census occupation data are based on ISCO 2008.

Comparability — Guatemala [top]

The 1964-1981 samples use their own classification, which is very similar to the COTA system. The 1994 and 2002 samples both used the ISCO 1988 classification system.

Comparability — Guinea [top]

The age universe includes persons age 10 and older in the 1983 sample and persons age 6 and older in the 1996-2014 samples.

All three samples use different coding structures, resulting in significant OCCISCO differences from one sample to another. However, the classifications used in the 1996 and 2014 samples are both consistent with the major group distinctions in the ISCO classification.

Comparability — Haiti [top]

The 1982 sample used the ISCO 1968 classification system. The 2003 sample categorized occupation according the ISCO 1988 system, thus no recoding was necessary.

While the samples have the same universe, the reference period for employment is different across years.

Comparability — Honduras [top]

The 1961 sample uses the COTA 1960 classification. The 1974 and 1988 samples use a unique classification. The 2001 sample uses the ISCO 1988 system. The 2013 sample uses the ISCO 2008 system. The universe in 2001 and 2013 differs from the previous samples.

Comparability — Hungary [top]

The occupation data in the Hungary samples are comparable, as they follow some ISCO coding structure. In 1970-1990, occupations are coded using a national classification equivalent to ISCO 1988. The 2001 sample applies ISCO 1988 and the 2011 sample uses ISCO 2008.

Comparability — Indonesia [top]

The samples use the same basic underlying classification (ISCO-68) with similar universes, and are therefore highly comparable over time.

Comparability — Iran [top]

The 2006 and 2011 samples report the current occupation of employed persons aged 10 or more. The current occupation is coded using the ISCO classification. Armed forces and police officers were not differentiated, and this category is coded to armed forces.

Comparability — Ireland [top]

The universe changes significantly over time; it includes employed, and sometimes unemployed, persons in the 1971-1991 samples, and includes experienced unemployed persons in the 1996-2011 samples.

The Ireland samples use a number of different occupation classifications, none of them based on ISCO. The 1986-1991 samples are coded from the same underlying classification, as are the 1996-2006 samples. The 2011 sample uses a unique classification system and provides detail at the three-digit level.

The 2016 sample uses the same classification system as the 2011 sample, but only provides detail at the one-digit level. Therefore, it is incompatible with the one-digit ISCO classification and is not integrated in OCCISCO. The occupation groupings for the 2016 sample are preserved in the unharmonized source variable.

Comparability — Italy [top]

The census samples use the ISCO 1988 classification system, requiring no recoding.

Comparability — Jamaica [top]

All Jamaica samples use the ISCO classification; thus, apart from the universe differences the data are fully comparable.

Comparability — Jordan [top]

The sample used the ISCO classification system, requiring no recoding

Comparability — Kenya [top]

In 1989 and 2019, the question applied to employed persons, whether at work or not; both samples include specifically persons on leave or sick leave, and the 1989 sample also mentions persons working on family holdings.

NOTE: In 1989, only a small fraction of persons actually reported an occupation.

Comparability — Kyrgyzstan [top]

The sample used the ISCO classification system but includes some additional categories.

Cattlemen and subsidiary plot farmers are classified as general workers and together comprise more than one-third of the workforce.

Comparability — Lesotho [top]

The 1996 sample uses the ISCO 1968 while the 2006 sample uses the ISCO 1988 classification system. In the 1996 sample, professionals and technicians cannot be separated. The universe is slightly different between the two samples.

Comparability — Liberia [top]

The 1974 sample uses the ISCO 1968 classification system. The 2008 sample uses the ISCO 1988 classification system and did not require recoding. The universe differs slightly across samples in terms of minimum ages.

Comparability — Malawi [top]

The unharmonized source variables for the 1987-2008 samples use nearly identical classifications, while the 2018 underlying data uses a different coding structure and provides less detail.

The samples have age universe differences: 10 and older in 1987, 1998, and 2018, while 6 and older are considered in 2008. The 2018 sample considers only persons who worked in the prior week, while the 1987, 1998, and 2008 samples also comprise unemployed respondents.

Comparability — Malaysia [top]

The 1970-1991 samples used the same occupational classification, and are therefore coded consistently within OCCISCO. The 2001 sample used the ISCO classification system, requiring no recoding.

Comparability — Mali [top]

The samples use entirely different occupational classifications, all of which had to be recoded significantly to fit the ISCO system.

Comparability — Mauritius [top]

The universe is identical across the three samples. The 1990 and 2000 samples use the ISCO 1988 classification system, while the 2011 sample uses a coding structure that is consistent with the ISCO 2008 classification system.

Comparability — Mexico [top]

The Mexico samples use national classifications to code occupations. The 1960 and 1970 samples used their own classification systems. The 1990-2000 samples used the same classification, while the coding structure is also common across 2010-2015 and is based on ISCO 2008.

In 1970 the question applied to the occupation the person had during the previous year, 1969, regardless of their status at the time of the census. In 1990, 2000, 2010, and 2015 the question was addressed only to employed persons.

The underlying 1995 classification is not detailed, and the resulting OCCISCO coding is therefore crude.

The 2005 Mexico census did not include an occupation question.

Comparability — Mongolia [top]

In 2000, the occupational classification was based on ISCO. The 2010 and 2020 samples use the ISCO-2008 classification, and the data are comparable across.

Comparability — Morocco [top]

The 1982 sample uses a different underlying classification than the one employed in 1994 and 2004. The 2014 sample uses a similar classification as the 1994-2004 samples, but there are several categorical differences. The occupation groupings for the 2014 sample are preserved in the unharmonized source variable.

Comparability — Mozambique [top]

The 1997 and 2007 samples use the ISCO 1988 classification. The 2017 sample uses the ISCO 2008 classification system. There are minor differences in the universes across the three samples.

Comparability — Myanmar [top]

The 2014 samples uses the ISCO 1988 classification system.

Comparability — Nepal [top]

The Nepal samples use the ISCO classification system, requiring no recoding.

Comparability — Netherlands [top]

The sample used the ISCO classification system, requiring no recoding.

Comparability — Nicaragua [top]

The 1971 sample uses a different underlying classification. The 1995 and 2005 samples use ISCO.

Comparability — Pakistan [top]

The 1973 Pakistan sample required significant recoded to fit the ISCO classification.

Comparability — Palestine [top]

All three samples use the 1-digit ISCO classification system and are fully comparable, apart from minor universe differences.

Comparability — Panama [top]

The 2000 and 2010 samples used the ISCO classification system, requiring no recoding. The samples for previous years used differing classifications that had considerable detail.

Comparability — Papua New Guinea [top]

In 1980 and 1990, only urban workers are in the universe of respondents. All samples are compatible with the ISCO 1988 classification system.

Comparability — Paraguay [top]

The 2002 sample uses the ISCO 1988 classification. Earlier samples use differing classifications, but they conform fairly well to ISCO.

Comparability — Peru [top]

The data for Peru are fully comparable between samples and were organized along ISCO lines, requiring minimal modification.

Comparability — Philippines [top]

All samples used the ISCO classification system, requiring no recoding.

Comparability — Poland [top]

The 1978 and 1988 samples were not integrated because they were not compatible with the ISCO coding structure. See the occupation unharmonized source variables for these censuses.

The 2002 sample had significantly less response categories but the coding structure follows ISCO.

Comparability — Portugal [top]

The 1991-2011 samples used the ISCO classification system, requiring no recoding.

Comparability — Puerto Rico [top]

The Puerto Rico data are recoded from differing underlying classifications. The universe of respondents changes over time as well.

Comparability — Romania [top]

The universe changes from people who are employed or seeking work in 1992 to only those people with a job in 2002. In 2011, the universe changes back to include only persons who were either employed or seeking work.

The 1977 sample includes an occupation variable, but it is not compatible with OCCISCO.

Comparability — Rwanda [top]

The 2002 sample uses the ISCO 1988 classification while the 2012 uses the ISCO 2008 classification. No recoding is required.

Comparability — Saint Lucia [top]

The 1991 Saint Lucia sample uses the ISCO classification. Occupation data are not available in 1980.

Comparability — Senegal [top]

The 2002 and 2013 samples use the ISCO classification, requiring no modification. In 2013, the source variable refers specifically to "profession" and it is available for inactive persons besides persons employed and experienced unemployed. Thus, for higher comparability across samples, the inactive persons are excluded from the universe in OCCISCO for 2013, but their responses are preserved in the corresponding unharmonized source variable.

The 1988 Senegal classification is highly industrial in character and is a relatively poor fit for the ISCO system into which it is coded.

Comparability — Sierra Leone [top]

The data report the occupation of respondents who worked in the last 30 days for the 2004 sample and the past 12 months for the 2015 sample. Both samples use a 1-digit ISCO 1988 classification.

Comparability — Slovakia [top]

The 2001 sample uses the ISCO 1988 classification and the 2011 sample uses ISCO 1988.

The 1991 sample used a national classification that was highly industrial in character and does not allow making many of the distinctions necessary for coding into the ISCO system. Consequently, the 1991 sample is not included in OCCISCO. See the source variable or OCC for the detailed 1991 classification..

Comparability — Slovenia [top]

The sample used the ISCO general category coding scheme, requiring no recoding.

The universe for Slovenia includes employed persons. Although no lower age limit is explicitly specified, no one under age 15 has an occupation.

Comparability — South Africa [top]

The question was asked of employed persons in all census years, but the age universe of respondents differs.

All samples were originally classified into the ISCO system by the national statistical office, so they required no recoding for OCCISCO.

Comparability — Spain [top]

The underlying codes from which OCCISCO was derived differ for each sample, and none were designed on the ISCO model. The 1991 OCCISCO codes were derived from 20 occupational categories, making for an especially crude classification in that year. Retirees are included in the 1991 universe and excluded from the 1981, 2001, and 2011 samples. Therefore, there are significantly more persons with occupation responses in 1991 than in the other years.

Comparability — Sudan [top]

The sample uses the ISCO classification.

Comparability — Suriname [top]

The 2004 and 2012 samples are compatible with the ISCO 1988 classification system and are fully comparable except for a difference in the age universe. In both samples, armed forces occupations are grouped with technicians and associate professionals.

Comparability — Switzerland [top]

All Switzerland samples use the ISCO classification. They each also have alternative occupation data available in the unharmonized source variables.

Comparability — Tanzania [top]

The data for all the Tanzania samples are based in the ISCO system, although the 1988 sample has no categories for technicians and operatives.

Comparability — Thailand [top]

The 2000 sample uses the ISCO classification, and 1980-1990 were similar, requiring limited recoding. The 1970 classification is considerably different.

Comparability — Togo [top]

The 1960 and 1970 samples use their own classifications but they are comparable. The 2010 sample was coded into the general ISCO classification, requiring no modification.

Comparability — Trinidad and Tobago [top]

The Trinidad and Tobago 1990 and 2000 sample use a classification compatible with ISCO 1988. The 1980 sample differs and it was not possible to distinguish technicians from professionals.

Comparability — Turkey [top]

All samples use the same classification and are therefore highly comparable with each other.

Comparability — Uganda [top]

The 1991 and 2002 samples use the ISCO classification system, requiring no recoding. There is a significant increase in skilled agricultural workers from 1991 to 2002, and a decrease in elementary workers.

The 2014 sample uses a national classification that is not compatible with the ISCO coding structure, so it is not integrated in OCCISCO. The occupation groupings for the 2014 sample are preserved in the unharmonized source variable.

Comparability — United Kingdom [top]

The 2001 sample used the ISCO classification system, requiring no recoding. The 1961, 1971 and 1991 samples used a different classification, but it generally made distinctions that facilitated recoding into the ISCO system.

Comparability — United States [top]

OCCISCO is coded from the classification used in OCC95US, which integrates differing U.S. classifications into the 1950 U.S. census system. In 1980 and 2000, there were major reclassifications of the underlying U.S. data, and comparisons spanning those years will be less robust. The U.S. samples are otherwise comparable, aside from the differences in the universe with respect to age, armed forces, and experienced workers not in the labor force.

Comparability — Uruguay [top]

The 1963 and 1975 samples each use a unique classification scheme. The proportion of technicians is smaller in 1963 with respect to other years, as most of these occupations are included within the corresponding professional categories. The 1985 sample uses COTA 1970. The 1996 and 2006 samples use ISCO.

The universe differs between 1963 and later years.

Comparability — Venezuela [top]

The universe differs across samples. The original data for 2001 were coded into the ISCO system, requiring no recoding. The 1981 and 1990 samples used the same classification. 1990 has more skilled agricultural workers than the surrounding samples.

OCCISCO is not available for 1971 because the documentation did match the values in the data.

Comparability — Vietnam [top]

The Vietnam 1999 census classified occupations using the ISCO 1988 system. The question applied to those persons who worked at least six of the previous twelve months. The 2009 sample used a three-digit occupation variable in the original data that was three-digits, but with the first digit corresponding to the ISCO one-digit codes. The universe for 2009 is includes persons age 15 and older who were at work, on salary, or had a job to return to.

The Vietnam samples originally coded the great majority of agricultural workers to the unskilled category. For purposes of comparability with other countries, these cases are coded to skilled agricultural workers in OCCISCO.

The census of 1989 used a different scheme that is impossible to fit in ISCO. See OCC for the codes.

Comparability — Zambia [top]

The 1990 and 2000 samples use the ISCO 1968 classification system; some recoding was necessary. The 2010 sample uses the ISCO 1988 classification system. The universe differs slightly across samples.

Comparability — Zimbabwe [top]

The 2012 sample uses the ISCO 1988 system, so no recoding was needed.