Marden Campos, OLAC

When we think about undercount problems in the census, we usually refer to individuals who could not be reached by the interviewers, either because their homes were not found or because, despite finding their residences, the interviews could not be conducted. Another common undercount case takes place when some individuals are forgotten by the informant, usually children, elderly and disabled people.

Nevertheless, a kind of phenomenon that affects significantly the census coverage is the failure to consider certain types of individuals as residents of the households. We consider these individuals as «marginal cases», as they depart from the concept of «typical citizen» that the census is expecting to find. Examples of such individuals are some indigenous peoples, homeless, undocumented migrants or the gypsies, a typical itinerant  people with «inaccurate» residence condition.

During the design of the census operations such groups are often treated as marginal cases, which would not affect the census operation. This situation restrict some types of analysis based on census data though, as I will try to discuss throughout the text.

Methodologically speaking, the most important concepts of the census are those related to units of collection and research. The first ones refer to the type of business or social entity that will be the object of the census registration (i.e. the places where the information will be collected). It may be represented by households and dwellings, firms, accounting offices or farms. The research units are the ones we want to obtain information from, which may or may not match with the collecting units, or may be represented by elements therein included. In the case of economic census, the collection units are the places where the information of the companies are stored, often different from research units. In the case of population census, the collection units are generally households and research units are households and individuals living them. In some cases, as in the 2010 Brazilian Census, information on the households surroundings was also collected.

We realize how important is the establishment of well-defined concepts of household and residents to carry out a population census. These concepts have been defined based on confrontation – not always peaceful – among institute’s experience in each country versus  international recommendations for conducting censuses.

Cognitively, these concepts are based on a notion of an «average person» and an «average household» of each country.  Individual slightly different from these average types can also be framed in the census, but we have problems with big detachment. For instance: by doing it in households, it is necessary that the country’s population have been living for some time in these households and, most importantly, recognize themselves as such. Again, households that depart a lot from the average type of country residence usually have coverage problems. While it is possible to incorporate residence options in the census questionnaires (such as the makeshift homes or hollow and huts in the Brazilian census), it is very difficult for the interviewer to find all individuals residing in shops, dens or even in tents and shacks. Similarly, street populations pose a serious problem for census operations in some countries as well as people who have two or more households.

In the same way, individuals with an indefinite housing condition or lack of legal permission for residence in a certain place are obstacles for the census enumeration. This is the case, for example, of the temporary workers and the international migrants. The fact that they do not have a permanent residence status either increase the probability of interview refusal.

In most cases, the size of the population groups that not fit in the census definitions is very small and does not spoil the results for large population aggregates. It is a fact that population censuses present limitations for studies of these populations, such as indigenous people and international immigrants. Unfortunately, policies drawn from the census probably exclude individuals from these populations, which is an incredibly serious problem.

Nowadays, if we aim to build a better knowledge about these individuals and their situation, we need to combine the census analysis with other sources of information, such as localized surveys and administrative records or else perform targeted sampling survey.

So, we must discuss our views on the notion of average individual that guide our statistics, our laws and our educational systems thinking if they really reach the real people that we want to know.