Comparability — Botswana [top]

The 2001 and 2011 samples include more response options than previous samples.

Comparability — Chile [top]

Most responses in 1970 are included in "zinc or tin", because that was the first material named, but the category includes other materials: "zinc sheet, roof tile, etc."

Comparability — Costa Rica [top]

The variable is largely comparable across years. Asbestos was added in 1973 and remained, clay tile dropped out in 2000, but waste material was added.

Comparability — Dominican Republic [top]

The universe differs slightly across samples. The 1981 sample has the category of "tiles," which is not included in later samples.

Comparability — El Salvador [top]

The 2007 census adds a response category for fiber cement sheets.

Comparability — Ethiopia [top]

The universe differs significantly across samples, including only urban households in 1984 and only households receiving the long questionnaire in 2007.

The 1994 integrated variable was made using two unharmonized variables- one for rural households and one for urban households. Asbestos roofing was omitted as a category from the 1994 sample. "Bamboo or reed" was added as a category for the 1994 and 2007 samples.

Comparability — Guatemala [top]

The data are comparable across samples. The "thatch" category includes straw, palm leaves, and similar materials.

Comparability — Honduras [top]

The 1961 census is a sample of individuals, not households. Information is only available for households consisting of a head.

Comparability — Indonesia [top]

The 1990 and 1995 censuses separate sheet metal into separate categories for zinc and asbestos, whereas asbestos is combined with zinc in other years.

Comparability — Jamaica [top]

Roofing material was not enumerated in the 1982 sample. The 1991 and 2001 samples are comparable in both categories and universe statement.

Comparability — Kenya [top]

The data are comparable across samples. The grass/makuti category from the 1989 census sample is split into two separate categories in latter samples.

Comparability — Malawi [top]

ROOF is not present in the 1998 sample. The 1987 sample has four variables for ROOF: one for each dwelling a respondent may own; this variable records roof materials of the first dwelling. Iron and tile are combined in both samples.

Comparability — Mali [top]

Tile roofs are included with sheet metal in 1987. In the 2009 sample, the unharmonized source variable differentiates roofs of sheet metal only and those that have sheet metal and a ceiling.

Comparability — Mexico [top]

In 2010, 2015, and 2020, the option "flat roof with wooden beams" is classified as "wood and other plant material".

Comparability — Morocco [top]

The universe for 1982 and 2014 samples is "all households" while the universe for the 1994 sample is "non-tent households".

Comparability — Mozambique [top]

The categories are consistent across years.

Comparability — Paraguay [top]

Most of the categories are consistent across years, with some variation.

Comparability — Philippines [top]

The category "Wood" includes bamboo in the 1990-2010 samples.

Comparability — Senegal [top]

The universe for the 2013 sample is "ordinary households" while it is "all households" for the 1988 and 2002 samples.

Comparability — Sierra Leone [top]

The categories are consistent across years.

Comparability — Tanzania [top]

In 2012 there is a response category for people living in tents, which is categorized as "other".

Comparability — Togo [top]

The 2010 unharmonized source variable provides more detail on type of sheet metal and type of thatch.

Comparability — Uganda [top]

The universe for the 1991 and 2002 samples is "private households" while it is "all households" for the 2014 sample.

Comparability — Uruguay [top]

The unharmonized source variables distinguish masonry with or without a "flat roof"; the 2006 census further distinguishes concrete with or without "protection", which can be disaggregated using the source variable.

Comparability — Zambia [top]

Most of the categories are consistent across years. However, the 2010 sample added several categories that were not included in earlier samples.