Housing Variables in Colombia 1964-2001
Susan De Vos
Department of Sociology
University of Wisconsin-Madison
Draft: December 1999
The purpose of this paper is to provide an overview of Colombia’s household censuses between the years 1964 and 2001, times for which there exist computerized microdata. The census for 2001 will include Colombia’s sixth housing census. It is reasonable to try to put the various censuses together for a time series of information, to integrate them into a harmonious format. There are problems with this of course, emanating from the facts that there is no standard definition of the household, no standard way to define the population that is to be enumerated, no standard set of housing items that are to be enumerated and no standard way to actually enumerate the same item. Colombian censuses over the last few decades provide excellent examples of each of these problems. Nonetheless, the problems pale in the face of the potential value of making reasonable comparisons. The data can provide such valuable insights, especially when combined with use of the corresponding population censuses, that the problems hardly outweigh the benefits of comparison. It is with the idea that useful comparisons can be made that I present this paper.
The paper has three parts. In the first part, there is discussion of trying to compare the censuses including a discussion of U.N. recommendations. In the second part, we actually provide time series data on a number of items, including some that have been used specifically to help indicate poverty in Colombia. We end on the more traditional note in family/household demography of viewing the lack of change in such conventional household measures as the Adults per Household ratio (AH) and Age Specific Headship Rates (ASHR).
Three problems with any attempt to compare housing data from two or more censuses or surveys are 1) how the household is defined; 2) whether the census is de jure (or de derecho) or de facto (or de hecho); and 3) what household items are enumerated. First, there are potentially important differences between buildings (edificios), residential buildings, living quarters or dwellings (viviendas), households (hogares) and families (familias). Most housing censuses are only concerned with residential buildings or dwellings and households. For instance, in documentation for the 2001 census (Anexos pp. 36, 37), there is a glossary of sorts that defines an edificacion as
Es toda construcción independiente y separada. Independiente quiere decir que tiene acceso desde la vía pública. Separada quiere decir que tiene paredes que permiten diferenciarla de otras edificaciones. Una edificación puede tener varias entradas y generalmente está cubierta por un techo. Por ejemplo: una casa, un edificio de oficinas o apartamentos, un hospital, un centro comercial, una vivienda construida con latas y desechos.
However the vivienda
Es la unidad habitada o destinada a ser habitada por una o más personas. Es necesario recordar que toda unidad tiene que ser independiente y separada ...
Censuses usually identify viviendas and gather pertinent information regarding them. For instance, the 2001 census will record 1) housing type, 2) condition of occupation, 3) predominant material of exterior walls, 4) predominant material of the floors, and 5) access to various services (electricity, piped water, piped sewage, natural gas, telephone, garbage collection). Then, within any residential building or vivienda there may be multiple households comprised of people who may or may not be related. It is this that is referred to as the household or hogar (Anexos p.37):
Es una persona o grupo de personas parientes o no, que viven bajo un mismo techo y generalmente comparten los alimentos.
Items usually reserved for the household include tenancy, number of rooms (excluding bathroom, kitchen and/or garage), restroom facilities and its cooking facilities. For example, does the household use a toilet with water or does it use a latrine? Are these facilities shared with other households or are they for the sole use of household members?
If people are related by blood, marriage or adoption, they are usually considered members of the same family but most censuses are not directly concerned with that. The 1964 census may have confused familia and hogar but it did provide the reader with a clear definition of the household. Subsequent censuses have been clearer that they are enumerating hogares rather than familias. That is, the household is a social group that occupies a building or physical dwelling, and it usually shares food and/or a budget, but its members need not be related. There is a problem however, because it may be generally agreed that the family and household are not the same but there is not as yet a standard definition of the household. In some cases, the household is simply equated with the people who live in the same living quarters. In other cases the "household" is the group who share food or a common budget. In the 1973, 1993 and 2001 censuses, it presumably is both.
In Europe, this divergence is often considered in terms of a ‘household-dwelling’ unit and a ‘housekeeping’ unit, the latter being the recommended unit.
The 1985 census was confusing because it did not list anything by household, instead limiting itself to the vivienda only. But it used a curious definition of the vivienda similar to that of a ‘household-dwelling’ unit, the result being that when a post-enumeration survey used the methodology of the 1985 census, the number of viviendas and hogares was about the same but when the survey used a more standard methodology, the number of hogares was still about the same but the number of viviendas was quite a bit less; the hogar to vivienda ratio using the 1985 census method was 1.03 whereas the more standard procedure produced a ratio of 1.37 (Patiño and Alzate, 1986 pp. 72). That subsequently the 1993 and 2001 censuses enumerated residential buildings separate from households again suggests that the 1985 experiment proved unsatisfying, but we are in the awkward position of having to continue using the anomalous data.
Perhaps the most confusing feature of the "housing" part of the census is that, although data are theoretically reported separately for the vivienda and the hogar, the samples that we use have only two records according to whether the item belongs to the housing census or the population census. Records of the housing census correspond to the vivienda and can be examined with that in mind, but to mesh housing information with information on the people living in the vivienda one must use population information recorded by hogar. One wonders why so much is made of the distinction between the vivienda and the hogar if the distinction cannot be used in most analyses. In a way this is to our advantage because the situation enhances the reasonableness of just comparing hogar data in 1964, 1973, 1993, and 2001 with the vivienda data in 1985.
De jure or de facto?
Although Colombia’s 2001 census will be de jure, its 1973 and 1964 censuses were de facto. The switch was apparently made to de jure in 1985 and appears to be in accord with current UN recommendations (1998:51):
With the growing interest in information on households and families and on internal migration, it is becoming increasingly desirable to prepare tabulations on the basis of usual residence rather than on place where present.... ... a usual-residence distribution is likely to be more accurate than a present-in-area distribution ...
Yet as far as I can see, no analysis was made of the implications this might have for household items. An article that carefully discussed a variety of aspects of the population and housing census was limited to merely saying that in general there did not seem to be much difference either nationally or locally when one or another definition was used (Vélez, 1986: 30). In most cases it should not matter since dwelling material will be the same, but there could be a difference in density or the number of people per room on the microlevel, and in ‘relationship to household head’ information and thus household composition.
The UN recommendations do seem to consider this possibility (1998: 52):
If it is also desired to obtain information on both the usually resident population and the present-in-area population, then ... A clear distinction must then be made ... among (a) persons usually resident and present on the day of the census, (b) persons usually resident but temporarily absent on the day of the census and (c) persons not usually resident but temporarily present on the day of the census.
Unfortunately, the Colombian censuses do not seem to have done the necessary enumerations to assess ‘usual’ and ‘temporarily present’ or ‘temporarily absent’ household members. As a result, it is impossible to assess how the definition change affected the results. For instance, we will see that the proportion of the household population listed as unrelated to the household head declined from 1973 to 1985 and 1993, but it is unclear how much of that decline was simply an artifact of enumeration differences.
Housing Items in the CensusThere is no standard set of items to be included in a household census. 1980 U.N. recommendations were as follows (1980:231):
Building in which living quarters are located (edificio)- characteristics of
Living quarters (vivienda)- characteristics and facilities of
Occupants of living quarters - number and characteristics of
Type of activity
Revised recommendations that came out in 1998 added a few more items: in addition to listing number of rooms it suggested enumerating useful and/or living floor space (UN 1998: 97-98). In addition to listing type of lighting or availability of electricity, it suggested listing type of solid waste disposal. In addition to listing number of occupants, it suggests listing occupancy by one or more households. In addition to listing tenure, it suggested listing rental and owner-occupied housing costs. On the other hand, the recommendations dropped items #14 and #16.
It appears that in Colombia, the first item (on type) is only enumerated as item #5 for the vivienda, not the edificio, except in the 1964 census. The third item on period of construction, considered "suggested" rather than "recommended," has not been included in any of the censuses. The most puzzling of the first set of items however, is item #2 on material of external walls. Latin American censuses have often listed housing material for three different building features, material of roof and floor as well as the material of exterior walls. In Colombia however, only the 1964 and 1973 censuses included an item for roof material while the type of floor is recorded in all the censuses. The 1964 census considered material a characteristic of the building but the other censuses considered it a characteristic of the vivienda.
The first few items for "living quarters," #4-#6, are reported for the vivienda rather than the hogar. Herein lies some confusion because in the Colombian censuses some questions related to the physical dwelling refer to the vivienda but others refer to the hogar while the UN refers to "living quarters" ("housing units and collective living quarters"). Thus access to services such as piped water, piped sewerage and electricity are usually reported for the vivienda, and the 1993 census added telephone service and garbage collection while the 2001 census additionally added an item about natural gas. Yet the details regarding toilet facilities, water facilities, cooking facilities, ownership, number of rooms and number of occupants are usually asked for each household within the building. The following table tries to outline the situation.
Table 1. Vivienda variables in existence in various censuses
Although it may not be appropriate for the UN to recommend it, enumerating household amenities such as ownership of a washing machine, a phonograph/tape/CD player, an automobile, a refrigerator or some other item that has particular significance could be very indicative of relative well-being and is usually reported for the household rather than an individual. The 1992 census for Chile provides an excellent example of the kind of items that might be included. Since it is often very difficult for many countries to talk about income or wealth in a simple monetized fashion, listing such amenities may be very informative. The 1964 Colombian census enumerated whether households had a radio or television. Those or related questions were not included in the 1973 census, perhaps because the answers were poor but neither was anything else tried. The result is that we have no income information nor amenity information aside from the limited items discussed further here. Probably in part due to this situation, Colombia has conducted a separate housing survey. While this may seem more cost-effective in certain respects, one cannot come close to having the coverage or sample sizes available from a census. One cannot juxtapose the wealth of information contained in the population part of the census with a potentially richer array of items in a housing part of a census.
Since the UN’s listing of original (1980) items #14-#16 relating to the conjugal family nucleus or household issues was probably the weakest part of the recommendations, it is not surprising that they were changed in the subsequent set of recommendations of 1998. For instance, instead of a focus on the conjugal bond of original item #14, some censuses should probably be more oriented toward capturing various aspects of a consanguinially-based kinship system (by blood rather than marriage). How is one to code a "queen bee" setup? How is one to consider a reincorporation of elderly family member(s) into the household? Regarding original and present item #15 (or #18) that lists several demographic characteristics of the head usually reserved for inclusion in the population part of a census, why are these particular variables selected (age, sex, activity status, occupation)? One could easily make a case for the inclusion of marital status, education and fertility from among variables in the population census. Nor is it clear why information should be limited to the household head. For instance, counting the number of household members who were active in the labor force could be a powerful piece of information. To assess household poverty, people sometimes monitor the school enrolment of children under 14 or 16 years of age (see e.g. DANE, 1989).
The 1980 item #16, "household," is the major household component of a population census. There is question about the relationship of household members to a designated "head" or reference person, sometimes augmented by a household composition variable. Regarding the question about relationship, there is no agreed-upon standard for designating one household member the head or reference person. In theory, any household member could be the head but the United Nations for a while seemed to continue recommending that this be the household’s oldest male even in the face of growing criticism of such a stance (1988). In the most recent set of recommendations the UN finally seems to have acknowledged that there are different views on who might be considered "head," that the actual internal dynamics of a household or the legal status of its members is unclear and that designating anyone as head may be inappropriate.
Colombian censuses do list the reference person as the head, but do not provide a clear definition of how the head is chosen. One works under the assumption that the "head" is whoever is designated as such by whoever answers the census questionnaire. As societal attitudes toward gender and/or age change, we may expect the "head" designation to change, perhaps subtly, causing pause in straightforward comparison over time and/or location of presumably factual data. At present, many countries have had a movement away from designating a head, instead designating someone a "householder," "household reference person" or some similar appellation.
The UN recommends listing the following relationships to the head or reference person (1998: 66): spouse (or companion), child (in-law), grandchild or great-grandchild, parent (in-law), other relative, domestic employee and other person not related to the head or other reference person. That is a fairly short list, and censuses often list more categories such as sibling, grandparent, aunt/uncle, or niece/nephew. In Colombia, the 1964 census had separate categories for boarder or lodger (inquilino)and child(ren) of servants. The 1973 census had a separate category for pensioner (but not for boarder). Subsequent censuses appear to have been limited to the short list but so doing may ultimately prove to be unfortunate when examining the situation of special segments of the population such as single mothers or elderly widows. This is discussed again in a future section.
The 1990 Mexican census had a reasonable approach toward listing ‘relationship to head’ potentially worth emulating: It enumerated a short list of relationship codes in a first column of what could be a fairly simple or fairly detailed, one-column or three-column variable. A researcher can thus either use the short list or the more detailed one. For instance, the short list has one category for "other relative" (6). In more detailed form, the "other relative" can be a father, mother, father-in-law, mother-in-law, sibling, half-sibling, sibling-in-law, grandparent, niece/newphew, aunt/uncle, cousin, and so on. Since it is customary in countries such as Colombia for extended kin to share housing, such information can be very important.
Relation information in the population census can be combined with information on number of people in the household and the age, sex and marital status of the various household members to construct a household composition variable. Is this item #15 again? The UN recommends a comparative composition variable that has four basic categories and many subcategories (UN, 1998:67-68):
Many frequency distributions that have ostensibly used this composition standard only use the four broad categories of one-person, nuclear, extended and composite household (e.g., UN Demographic Yearbook of 1984). This separates out any household with an unrelated member from the rest of the household and, indeed, makes the unrelated membership more important than the nature of the family core. That is, a household with an extended family is categorized as ‘composite’ if it also has a boarder; similarly a nuclear family is categorized as ‘composite’ if it also has a boarder. In the simplified version no distinction is made between basically nuclear and basically extended family households as both are simply considered composite.
Alternately, an anthropologist and historian together devised another scheme that did focus on the family core of (Hammel and Laslett, 1974). Motivated by studies of the household in historical Europe, their scheme has five basic categories: 1) solitaries (or one-person), 2) no family, 3) simple family, 4) extended family, and 5) multiple family. The fifth category might be included by many into the fourth category because it involves the co-residence of more than one nuclear unit. It was designed particularly with the frérèche of Yugoslavia (when two married brothers coreside) in mind. The second ‘no family’ category includes coresident siblings, coresident relations of other kinds or persons not evidently related. The simple family household includes a) a married couple living alone, b) a married couple living with child(ren), c) a widower with child(ren), or a d) widow with child(ren). (Marriage does not have to be formal as the marital unit in actuality is the ‘conjugal’ unit which includes a consensual union.) Unrelated household members are not a major part of this scheme, perhaps because of the common residence of boarders and lodgers in historical European households. Thus extended family households with an unrelated household member would still be categorized as ‘extended’ while simple family households with an unrelated household member would still be categorized as ‘simple’ in a version with only five (or four) categories. Only a more detailed version would differentiate between households with or without unrelated members.
No Colombian census has manufactured a household composition variable except in-house. Fortunately, the University of Wisconsin-Madison’s Changing Household in Latin America project did manufacture those variables in sample microdata for 1973 and 1985 with the following result:
Comparing the two sets of figures, it appears that the proportion of households comprised of only one person was roughly the same at both times. They also suggest that the ‘simple’ and ‘complex’ proportions of households were also about the same. In both 1973 and 1985 it seems, a little more than half the composite households had a ‘simple’ family core, and that rather than being less than half of all households they actually constituted a majority of the households.
In this section, we deal with various items mentioned above, and try to suggest how they might be compared in different censuses despite their different wording. Some of these items help indicate basic adequacy of the living quarters. Others refer to basic services such as piped drinking water, sewers, electricity and most recently telephone, garbage collection and natural gas service. There are other characteristics that we simply refer to as housing characteristics such as the existence of a toilet or kitchen. Colombia’s housing census covers tenancy as well as crowding. Since one of the uses of housing indicators is to help indicate poverty, we try to follow previous work using housing information to show the pervasiveness of poverty in Colombia, and how this has changed, but not as much as could have been hoped, between 1973 and 1993.
Household typeFirst, we need to assess the type of household where we deal with type of living quarters. Conceptually, there appear to be four different types, with a fifth type of ‘other’: house, apartment, room and shack. The trouble is that the 1993 census only considered three types, apparently including the ‘shack’ category in the ‘apartment’ category. Furthermore, although the 1985 census ostensibly considered 8 types including shacks, the microfile sample from the 1985 long form in actuality only considered two types, houses and apartments. Consider the following:
How does one interpret this? My inclination is to compare samples limited to houses and apartments in 1973 with the 1985 data, or compare 1973 directly with 1993 figuring that ‘shack’ in 1993 was included as ‘apartment.’ Doing this would lead one to surmise that the proportion of households living in a house as opposed to some other arrangement has stayed about the same. I do not see a reasonable way to compare the 1985 data with the 1993 data without acknowledging the important difference in the sample representation. On the other hand I see figures, such as those on poverty reproduced in a 1989 publication, suggesting that bias in the 1985 sample may not be all that great compared to that of the others.
Exterior Wall MaterialA second basic characteristic of a living quarters is the type of material used in its construction, and a basic indicator of that is the material of its walls. Confusingly, this can be quite a variety ranging from highly durable brick to vegetable material such as bark to throwaways. And each census has made its own categories. Attempted standardization by durability can take on four categories (Arias and De Vos, 1996: 62). Here, we get:
But we must note that pressed dirt has to be included in (1) in 1985 and again in 2001. It is currently impossible to make a true standard for this variable.
Floor MaterialAlthough not included in the UN recommendations, most Latin American censuses record a dwelling’s floor material as well as its wall material. There are three basic kinds of floor material: 1) dirt, 2) cane or some impermanent vegetable matter, and 3) a more stylized covering such as brick, vinyl or rug. Even so, each census coded flooring somewhat differently. The following is an attempt to show standardized distributions:
Especially problematic to me is how to classify "madera" since it could indicate nice flooring or be rather impermanent. For instance, the 1973 census lists "madera" but later censuses only list "madera burda".
Access to basic servicesIn 1980 the UN recommendations listed piped drinking water, sewerage and electricity as basic services. In 1998 it added telephone, garbage collection and natural gas service but we only can cover the first three items in the 1973 to 1993 period:
It would seem that piped water generally became more commonly available between 1973 and 1993 but that access to a sewage system stayed fairly limited. While electricity could be considered a basic service (e.g. Arias and De Vos, 1996), it is often not considered basic to health the way access to piped water and a sewage system is. On the other hand, one might consider bathing facilities basic to good hygiene yet only the 1964 census reported the existence of bathing facilities.
As we will see, one measure of poverty is if either piped water or sewerage service was lacking in urban areas or if one of these was lacking in rural areas. With this measure, the proportion of households in poverty declined nationally from over 45 percent in 1973 to 27 percent in 1993. But even with the more lax demands on rural living, the proportion not meeting the poverty requirement was high, 80 percent in 1973 and 57 percent in 1993. As might be surmised from the overall figures, the lack of a sewage hook up was more prevalent than the lack of piped water in both urban and rural areas.
Other basic household qualitiesThe UN recommendations list items such as drinking (or cooking) water, toilet facilities, bathing facilities and cooking facilities as ones to help describe basic household as opposed to residential building facilities. The Colombian censuses consider these characteristics of the hogar instead of the vivienda (except in 1985). Corresponding indicators are not standardized however, in part because people can have septic tanks if they are not hooked up to a sewage system and they can cook in a room that is not designated solely for the purpose of cooking. Simple distributions of our samples according to these items are:
It would appear that having access to a toilet with running water was much more common in 1993 than in 1973 and that it was also more common that the toilet not be shared. Again, this was much more the case in urban than in rural areas. In 1993 in fact, still almost half of all rural households had no toilet at all, even when such facilities as a latrine are counted.
The question on access to basic water service is only augmented somewhat by the more detailed description of from what source the water comes because almost 70 percent in 1973 and 80 percent in 1993 of the households obtained its water from a pipe or aqueduct. Still, it is useful to note that this increase was due to a shared decline in obtaining water from a well or a river/stream/spring. These latter two water sources continued to serve a substantial proportion of the households in 1993.
The question about a kitchen, which kept changing and was eventually dropped by the 2001 census is rather more confusing. The 1985 census used two questions, one of which was similar to the question asked in 1973 (about sharing the facility) and the other of which was similar to the question asked in 1993 (about where cooking was done). It would seem that having a kitchen may have become slightly more common between 1973 and 1993 but that how cooking was done proved too complicated to code adequately.
TenancyAnalysts are often curious about whether people own their dwellings or not for a whole host of reasons such as having greater commitment to the area, having pride in place or having something to pass on to children. Yet tenancy seems to be a tricky variable. For instance it does not seem related to the type of building material used or the tenants’ access to basic services (Arias and De Vos, 1996). Tenancy is often not used to help assess poverty. Still, it is a basic household attribute that is almost always recorded. There appear to be three basic types of reportable tenancy, 1) owning (whether still in the process of paying or not), 2) renting (or subletting), and 3) other (such as usufruct). The Colombian censuses do not assign tenancy to any particular individual or subset of individuals but rather consider it a characteristic of the entire household. This can be confusing when combined with the tenuous nature of assigning headship but may also be more realistic than anything else. Below are the distributions we find in our 1973, 1985 and 1993 samples:
It is for distributions such as these that I tend to be somewhat skeptical of the comparability of the 1985 figures because, if taken on face value, the figures suggest that ownership increased for a while during the 1970s but decreased again in the 1980s and early 1990s. Conversely, renting rather than owning decreased and then increased? What appears most clear to me is that some other form of tenancy beside ownership or renting declined dramatically after 1973.
Number of rooms, household size, and densityThe U.N. recommends that censuses report a household’s number of rooms excluding any kitchen or bathroom (or garage). This can be confusing if enumerators are not careful because some households may allocate a room (or more) as a work room rather than a living room. (Some censuses have recently taken to measuring the amount of living space, not just number of rooms, but Colombia does not seem to do that yet.) Although most households contain between one and five rooms, some households may have as many as 15 or more. Similarly, although most households contain between 3 and 10 people, some may have 20 or more. Together, the number of people divided by the number of rooms indicates density, a measure often used to help indicate overcrowding and hence poverty. Here we report the number of rooms, the number of household members, and density in terms of means, standard deviations, minimums and maximums:
Density was reduced significantly in the 20 years between 1973 and 1993 from an average of more than 2.5 to about 1.7. The average number of rooms in a household increased slightly and the average number of people dropped by over one person from almost 6 to under 5! This is probably due largely to the drop in fertility from a TFR of about 6.8 children per women in the early 1960's to about 2.9 in the late 1980s (U.N., 1995).
A major use of information obtained from a housing census is to track the well-being of the population. An important study that examined poverty in Colombia in the mid 1980s with census data used three indicators constructed from housing data (DANE, 1989: 15-16): 1) households in inadequate living quarters, 2) households in living quarters without basic services, 3) crowding. It furthermore made a distinction between municipal heads and the rest of the country. Finally, it considered people with two or more indicators to be in misery, not just poor.
The large study examined different geographic areas and reported its findings in two volumes (DANE, 1989). Here, I merely show the percentage of households in poverty according to each indicator, and the percentage of households having two or more characteristics of poverty (said to be in misery) for 1973, 1985 and 1993:
These overall figures suggest that there has been a slow but steady decline in the basic poverty of Colombian household although basic poverty still afflicted over a third of the country’s households in 1993. Most dramatic was the decline in overcrowding (when 3 or more people to a room is used as the main definition).
What was particularly impressive was the difference between urban and rural poverty, poverty afflicting almost two-thirds of rural households in 1993:
A variable commonly found in the population census but referring to the household is "relationship to head."
Clearly, the proportion of the household population who are ‘children’ of the household head has declined between 1973 and 1993 and the proportion of actual heads or spouse/companions of the head has increased. The proportion who are grandchildren seems to be lower in 1993 than in the 1985 sample although higher than in 1973.
Additionally, the proportion of the household population listed as ‘other relative’ of the head is rather high, even if less in 1993 than in 1973. Based on my examination of some 1990 Mexican data on elderly women (De Vos, forthcoming) many of these ‘other relatives’ might be siblings and their children, or grandparents of the head. There is reason to suggest it worthwhile to identify such relatives more positively as they could constitute a larger proportion of the household population than children in-law. Also,
Ruggles’ careful assessment of the demography of the unrelated individual in the United States in the first half of the 1900s suggests that there are a host of reasons the proportions might be different in different censuses including issues surrounding definitions of households and whether students are enumerated at their parents’ home (1988).
Demographers often make a first appraisal of a country’s household structure by calculating simple measures using information on number of households, a population’s age distribution, and the age and sex of household heads. We make a similar first appraisal in this section with data for 1973, 1985 and 1993. There has been a decline in average household size and in the average number of adults (aged 15 and over) per household. Also during that period, headship has increased among both men and women at older ages. But most important of all perhaps is the observation that the headship rates observed for 1993 are not all that different from the ones calculated for 1985 or even 1973. What the situation might be over a longer period of time is unclear.
Mean Household Size and the Adult per Household Ratio
One of the major benefits of conducting household and population censuses simultaneously is that it is then possible to juxtapose demographic characteristics of household members with characteristics of the household. Even on the aggregate level, a fairly simple demographic indicator of household complexity is the adult per household ratio (AH). Early studies started examining household structure by calculating the mean household size (MHS) but it was soon found that even large differences could merely be due to fertility differences. Looking at the average number of adults per household (AH) could give a better indication of actual structural change (Burch, 1970). Here I show you both:
On the surface anyway, it would appear that households are not only becoming smaller, but that they also may be coming less complex. The former could be due principally to the drop in fertility but the latter is more indicative of simpler households. Still, an AH of 2.9 indicates an intermediary level of complexity compared to a low of perhaps 2.3 in the United States in 1970 or a high of 3.4 in Ireland in 1971 (De Vos, 1987).
Age Specific Headship Rates
As a summary measure of central tendency AH is quite informative but age specific headship rates, requiring more information are at once both summary measures and more indicative of a distribution. They are the proportion of any age and sex group heading a household, and are usually reported as percents. The following are age-specific rates for males and females in 1973, 1985 and 1993 (see also Figure 1).
Colombian ASHRs have the expected curves: Male rates are low at young ages, peak during late middle age and decline slightly in old age. Female rates are very low at age 15-24 but climb steadily to about a third for those aged 65 and above. They are somewhere between the typically high rates of North America and the typically low rates of East Asia. Thus they are consistent with the existence of many simple families but some complex family households as well. They are consistent with some older people heading their own, perhaps solitary household, but also with many older people residing with others.
Headship rates from 1973 to 1993 appear amazingly similar. Those for middle aged men hover around 80 at all three census times. Those for middle aged women climb from about a fifth to a third. The most significant change between 1973 and 1993 appears to be a somewhat increased tendency to head among older men and women, perhaps signifying increased solitary living (see Arias and Palloni, 1999). Among men aged 65 and older, headship increased from 72 to 79 percent. Among women aged 65 and over, headship increased from 32 to 39 percent.
The housing component of Colombia’s national censuses since 1964 have the commendable quality of having followed U.N. recommendations for censuses and surveys. This has made them largely comparable with each other and fortunately with many other Latin American censuses as well. In addition to covering most recommended items they have distinguished between household and family and between living quarter and household, and have used fairly common definitions. The major anomaly in this regard was in 1985 when the census tried using a novel and now-abandoned definition of the vivienda which in many ways was more similar to most definitions of the hogar than regular definitions of living quarters. Colombian censuses also switched from being de facto to de jure after 1973 to better catalogue migration and usual household arrangements. They appear to have ironed out many wrinkles in coding as the 2001 census will be very similar to the one conducted in 1993. One can easily see an evolution in categorization and wording since 1964 of course, and the differences can sometimes make proper interpretation of a time series difficult.
There is always room for improvement of course. Two fairly simple ways in which the Colombian census could be improved regarding its dealing with the household would be to 1) develop a more detailed list of ‘other relatives’ to include such categories as ‘sibling’ or ‘grandparent’ of the head, and to 2) develop a simple household composition variable such as the one recommended by the U.N. Mexico’s 1990 census provides excellent examples for both these points. A third improvement would be the coverage of additional "standard of living" items such as use/ownership of such items as a refrigerator, CD/tape player, motorcycle or automobile. Chile’s 1992 census provides a nice example of this.
The usefulness of a housing component to a national census is now well established and the 2001 census will contain Colombia’s sixth housing component. Even for demographers trained to use the population part of the census, housing information can be very useful. This is especially so in situations in which there is no or only a speculative variable on household income. Then, housing variables can be used to assess basic housing quality. Even a bare list of items can divulge important information: this paper followed past studies to develop three possible indicators of poverty and found the proportion of households with at least one indicator strikingly high. The list was rather stark however, and one could easily suggest that it be augmented to include more above-basic amenities.
Arias, Elizabeth and Susan De Vos. 1996. "Using Housing Items to Indicate Socioeconomic Status: Latin America." Social Indicators Research 38:53-80.
Arias, Elizabeth and Alberto Palloni. 1999. "Prevalence and Patterns of Female-Headed Households in Latin America:1970-1990." Journal of Comparative Family Studies 30(2):257-79.
Burch, Thomas K. 1970. "Some Demographic Determinants of Average Household Size: An Analytic Approach." Demography 7:61-70.
De Vos, Susan. 1987. "Latin American Households in Comparative Perspective." Population Studies 41:501-17.
________. forthcoming. "Kinship Ties and Solitary Living Among Unmarried Elderly Women in Chile and Mexico." Research on Aging.
Ekouevi, Koffi, Mohamed Ayad, Barrère, and David C. Cantor. 1991. "Household Structure from a Comparative Perspective." Chapter in Conference Proceedings. Pp. 1547-77 in Demographic and Health Surveys World Conference Proceedings Vol. 3. Washington, D.C.: Westinghouse.
Hammel, Eugene and Peter Laslett. 1974. "Comparing Household Structure Over Time and Between Cultures." Comparative Studies in Society and History 16:73-109.
Palloni, Alberto, Susan De Vos, Elizabeth Arias, Beth Fussell, and Hanta Rafalimanana. 1994. "Changes in Families and Households in Latin America, 1970-1990." Population Association of America Meetings.
Patiño, Ximena and María Cristina Alzate. 1986. "Definición y Tipología de Vivienda en el Censo 85." Pp. 41-82 in Seminario Taller Sobre el XV Censo National de Población y IV de Vivienda, 1985, edited by Carmen Elisa Florez Nieto and Rafael Echeverri Perico. Bogotá. Colombia: Departamento Nacional de Planeación.
Republic of Colombia. Departmento Administrativo Nacional de Estadistica AND UNICEF. 1989. La Pobreza en Colombia (Tomo I and II). Bogotá, D,E.
Republica de Colombia. DANE. 2000. "Propuesta General - Anexos." Final Version. sent to me by McCaa.
Republica de Colombia. Departmento Admnistrativo Nacional de Estadística. 1967. XIII CENSO NACIONAL DE POBLACION (15 DE JULIO DE 1964) RESUMEN GENERAL. Bogota, D.E.: Imprenta Nacional.
República de Colombia. Departmento Administrativo Nacional de Estadística. 1996. Resumen Nacional. XVI Censo Nacional de Población y V de Vivienda. Santafé de Bogotá, D.C.: República de Colombia.
Ruggles, Steven. 1988. "The Demography of the Unrelated Individual: 1900-1950." Demography 25(4):521-36.
United Nations. 1980. Principles and Recommendations for Population and Housing Censuses. New York: United Nations.
________. 1981. Estimates and Projections of the Number of Households by Country, 1975-2001. ESA/P/WP.73. New York: United Nations.
________. 1988. Improving Statistics and Indicators on Women Using Household Surveys. Studies in Methods Series F, vol. 48. New York: United Nations.
________. 1998. Principles and Recommendations for Population and Housing Censuses Revision 1. M, vol. 67/Rev.1. New York: United Nations.
United Nations Department for Economic and Social Information and Policy Analysis. 1995. World Population Prospects (The 1994 Revision). New York: United Nations.
United Nations. Department for Economic and Social Information and Policy Analysis. Population Division. 1995. World Urbanization Prospects: The 1994 Revision. New York: United Nations.
Vélez, Ignacio. 1986. "Presentación del XV Censo de Población y IV de Vivienda de Colombia -- Censo 1985." Pp. 29-40 in Seminario Taller Sobre el XV Censo National de Población y IV de Vivienda, 1985, edited by Carmen Elisa Florez Nieto and Rafael Echeverri Perico. Bogotá. Colombia: Departamento Nacional de Planeación.