The debate on the future of the population census: Relevance for South Africa?

ABSTRACT

The cancellation of the Canadian census long-form questionnaire in 2010 and its subsequent reinstatement in 2015 resulted in a vigorous international debate on the future of the mandatory census. Questions concerning state invasions of personal privacy, party political interference and public trust and cooperation in state data-gathering were raised. As a result, alternative forms of demographic data collection were explored and the experiences of other countries examined. National population registers and non-mandatory surveys offered alternatives, but each had their own distinct disadvantages. South Africa has continued to conduct mandatory detailed censuses since 1994 and the enumerations have encountered little political opposition because they are presented as essentially nation-building exercises providing credible and relevant data, upon which national policies may be based and monitored. At present there is little incentive to change the current format of statistical collection in the absence of a viable alternative.

KEYWORDS:

• South Africa
• population statistics
• census
• survey
• register

1. Introduction

The furore created by the Canadian government’s cancellation of the long-form questionnaire in the 2011 census has emphasised the importance of comprehensive population enumerations for national development (Dillon, 2010; Green & Milligan, 2010; Veall, 2010). The action was motivated by the libertarian populist character of the Conservative Party administration demonstrating its concern with state invasion of citizens’ privacy and suspicion of the degree of independence of the national statistical agency (Ramp & Harrison, 2012; Walton-Roberts et al., 2014). There was little reference to developments in data-gathering which had been unfolding in Europe, where viable alternatives to the traditional style of the census had been explored since the 1970s and adopted by several continental statistical services (Valente, 2014). However, in other continents there had been little change in the traditional form-led questionnaire enquiry as it had evolved from the nineteenth century (Kukutai et al., 2015). In large measure this reflected the different political and historical trajectories of the states concerned, notably in the population’s perception of the demands imposed by the periodic conduct of an official mandatory personal enquiry into many individual and household attributes and assets. Nevertheless, most national statistical services examine possible viable alternatives to what is one of the largest and most expensive government operations in peacetime. This article proposes to examine the development of the current state of census-taking leading to the Canadian crisis and to review the alternatives that have been implemented in those countries seeking a new form of population inventory. This is followed by a review of the situation in South Africa and a conclusion.

2. The development of the census

The census is necessarily a state enterprise owing to its expense and the need to enforce a mandatory response by means of the seldom enforced threat of sanction for non-compliance in the form of fines or even imprisonment (Coleman, 2012). Early European and North American censuses were essentially concerned with basic demographic information, notably with manpower, growth, occupation, household structure and often issues of identity: language, religion and race. However, by the mid-nineteenth century governments were also concerned with such social issues as public health and education (Higgs, 2004). The need to know more about each individual increased as the state assumed an expanding role in social welfare and planning ranging from old age pensions to highways. As a result the length of the census questionnaires increased. Thus in the United States the number of questions asked increased from 12 in 1850 to 30 in 1890 and to 53 in 2000 (US Census Bureau, 2002). The motive for this expansion was the concept that public policies needed to be based on relevant, reliable and accurate information and that the more statistics available to guide policy makers the better.

In many countries, and notably in the United States, the volume of information gathered increased dramatically with each census in the nineteenth and twentieth centuries. The costs of converting the original answers into tabular form for the use of the state bureaucracy and the wider public was time-consuming, and, even with the introduction of mechanical tabulators and later computers, labour intensive. A further problem was the concentration of work in the period immediately around the date of the actual enumeration. Developments in statistical theory permitted the introduction of sampling, with a substantial reduction in the volume of work (Kish, 1979). Thus from the 1960 census the United States introduced a two-tier questionnaire, with 20% of households having to fill in the full long-form questionnaire, while the remaining households filled in a shorter form, containing only a few basic questions. This resulted in a substantial reduction in the work involved and hence a speedier release of the results, without a significant reduction in the integrity of the statistics. Nevertheless, considerable time continued to be devoted to the post-enumeration verification process (Salvo & Lobo, 2013). This, however, did not prevent political wrangling over the figures in the subsequent apportionment of federal funds and congressional seats (Hannah, 2001). Significantly, Canada followed suit in instituting long-form and short-form questionnaires in the 1971 enumeration.

Although the government controls the census, it relies upon the active cooperation of the general population to make complete and accurate answers to the questions posed. Thus controversial questions, likely to incur hostility from some sections of the population and which might therefore impair the integrity of the results, are usually avoided. In this manner, the involvement of everyone in the country becomes a national enterprise. This was noted in Great Britain in the nineteenth century where all classes were treated equally and incorporated into the statistical nation (Higgs, 2004). The same reasoning was in evidence in post-colonial south-east Asia where the census was publicised as an inclusive nation-building exercise (Anderson, 2006). This sentiment was echoed during the Canadian crisis with the observation that ‘participation in statistical inquiries is an essential part of citizenship in a modern democracy’ (Dillon, 2010:398). Thus, at all stages of the census, public opinion is invoked and public consultations as to content are widespread. The recent consultation on the Canadian census thus received over 500 submissions (Statistics Canada, 2014). The utility of the results not only for government purposes, but for the wider public has been evident from the early stages of the census, as the detail was marshalled to support a wide variety of often contradictory arguments and to support a variety of programmes of action (Levitan, 2011). It might be noted that similar movements were evident in nineteenth-century colonial states as local elite groups recognised the possibilities offered by the census to further their own ends (Cohn, 1987).

Dissent was sometimes in evidence in compliance with the census demands, but the general consensus as to the benefits of the census remained intact until the late twentieth century. The rise of libertarian movements hostile to the state and suspicious of the motives behind personal data collection by the state resulted in increasing resistance to the perceived intrusive nature of the census in a number of countries. This was particularly evident in Western Europe where the authorities sought ways of circumventing the problems raised, through alternatives to the lengthy questionnaire (Hannah, 2009). Although resistance to the census was initially raised by left-wing anti-state and anti-capitalist movements, it later linked with neo-conservative movements, notably in North America and the United Kingdom. This found political support in right-wing political parties which gained power in the new millennium (Thompson, 2010). Similar resistance to the inquisitive nature of census-taking was encountered among populations which had gained freedom from dictatorships in the 1990s (Uehling, 2004; Hull, 2005).

In 2010 the Canadian Conservative Party government’s rejection of the detailed, long census questionnaire seized upon the libertarian populist objections to the census to claim widespread dissatisfaction with the state’s apparent invasion of personal privacy (Ramp & Harrison, 2012). The question on the number of bedrooms in each household raised most ministerial ire, although it was concerned with providing a measure of overcrowding and was retained in the subsequent voluntary survey (McDaniel & MacDonald, 2012). The argument proved to be spurious because only three complaints about the census had been filed with the Office of the Privacy Commissioner of Canada in the previous decade (Thompson, 2010). It has been suggested that the government motive was part of the attempt to reduce the independence of Statistics Canada and demonstrate political control by the state (Walton-Roberts et al., 2014). Notably, the minister responsible for the census stated with reference to the independence of Statistics Canada: ‘Sometimes some of them like to think they are – but that doesn’t make it so. They report to a minister’ (Chase, 2010). Yet although answering to a minister, the organisation relied upon a popular perception of political independence as well as the professional integrity of its staff for the retention of public trust in the accuracy and reliability of the statistics it produced. The poor response to the ensuing 2011 Canadian Household Survey failed to produce robust statistics to the standard of the former long-form census, with resultant problems for their users (Grant, 2015). The report by Statistics Canada in considering its options for the 2016 enumeration concluded ‘that the traditional census approach is the only viable approach for 2016’, and indeed ‘for 2021 and beyond’ (Statistics Canada, 2012:72). The re-election of a Liberal Party government in late 2015 led to the immediate reinstatement of the long-form questionnaire for the 2016 census, as a political commitment to ‘data-based decision-making’ (Editorial, 2015).

Governmental hostility to independent statistical collection rested upon the quest by the census officials for as complete an enumeration as possible, which usually involved the expensive follow-up of people marginal to society such as minority groups, the very poor, immigrants and the very rich. It was thought that they in general did not support the conservative political agenda or view of society, from which they could be excluded by non-enumeration (Beaud, 2012). Alarmingly it was observed that:

It would appear that some of our political leaders consider that government can be based on values and good ideas, without the need for data and research. This is not unlike the view in some social science circles, that we should move to a post-positivist approach that is based solely on theory and qualitative approaches rather than official large-scale state-produced data. (Walton-Roberts et al., 2014:36)

‘Decision-based evidence making’ has a long history in the private gathering of survey data by political ‘think tanks’ and others where questions are often framed in a manner to support pre-determined policies (Slater, 2014:957). The same is also possible in state-conducted censuses, as the post-independence history of census-taking in Nigeria attests with the problems of government interference and the politicisation and suppression of results, when an ‘absence of a reliable census data legitimately calls to question “the rationality of public policies”’ (Mimiko, 2006:2). Indeed poor quality and unsuitable statistics have hampered development projects in many African countries for similar reasons (Jerven, 2013). A substantial measure of independence in the collection of statistics, as free as possible from direct government intervention, remains essential for their integrity to be generally accepted on both sides of any policy argument.

3. Alternatives to the census

In the face of rising opposition, alternatives were sought to fill potential gaps in the national population statistics, which would result from the absence of a mandatory questionnaire-led census (Prévost & Lachapelle, 2012). These included the utilisation of national population registers and databases, the substitution of voluntary surveys for the compulsory census and a hybrid short-form census and supplementary more detailed voluntary survey. In the 2010 round of censuses, 17% of the 227 countries which conducted a population count used some form of alternative to reliance solely on a full questionnaire-based enumeration (Kukutai et al., 2015).

Many states, notably in Europe, already compile registers of their resident populations and seek to track the movements of non-residents and migrants (Szreter & Breckenridge, 2012). A large amount of data from other sources can be linked to the register, which can be effectively updated continuously. It might be noted that it is very difficult to live in most countries without some form of link to an official database (e.g. for social services, police and taxation) or to a commercial database, including bank and credit agencies, and being required to provide proof of identity, notably possession of a passport, identity card or birth certificate (Baffour et al., 2013). The simplest record is the register of births, marriages and deaths, which provides the minimal legal proof of existence. In fulfilling this task, registers have a fundamentally different purpose when compared with the census. Modern registers are concerned with the recognition of the individual, while censuses are concerned with the population as a whole, in which the identity of the individual is subsumed in the characteristics of the whole. The establishment of a national population register, however, is expensive, but its maintenance is relatively cheap and up-to-date information can be obtained at any time (Longva et al., 1998). Furthermore, the maintenance of a national register is part of the continuous bureaucratic function and therefore excites little public attention or opposition. However, the register is essentially an administrative database of individuals, and it is not possible to add new questions in the manner that censuses and surveys can adapt to changing circumstances (Prins, 2000). However, it is possible to merge several government registers in order to provide a wider range of information for the creation of a ‘virtual census’ (Statistics Netherlands, 2014). Usually such data are purely demographic and identity questions (language, religion and race) are usually excluded on account of their sensitivity. Their inclusion, it has been noted, could lead to making such divisions in society permanent and therefore ‘the racialization of society’, with detrimental effects on national unity (Coleman, 2012:348). This was most marked in France where ideological concerns for national unity have resulted in a deliberate ignorance of identity issues of race, religion and language even in the census (Simon, 2008). However, registers have been and are a means of government surveillance of the population, and may be perceived as intrusive. Thus Statistics Canada concluded that the use of registers ‘may raise issues related to the Canadian Charter of Rights and Freedoms’ and that the Privacy Commissioner of Canada had expressed her severe reservations (Statistics Canada, 2012:54).

The second alternative is the non-mandatory survey. The use of surveys, which the respondent does not have a legal obligation to answer, offers the possibilities of requesting a wide range of different pieces of information. However, it requires another means of obtaining a base figure for the whole population, as well as the comprehensive household address list, based on detailed mapping and fieldwork, as required for the conduct of the census. Thus in the 2011 Canadian census, the surviving short-form mandatory questionnaire elicited a 98% response rate, but the voluntary National Household Survey, which replaced the long-form census, elicited only a 69% response (Walton-Roberts et al., 2014:41). Because there are differential response rates to survey questionnaires by different groups, some are under-represented and therefore ‘hidden’. Thus it was noted that, ‘Even if the long-form is reinstated in future years, 2011 will always be a black hole of absent self-knowledge’ (McDaniel & MacDonald, 2012:268). This raises the initial concern expressed by the Chief Statistician of Canada in his resignation letter over the abolition of the long-form census: ‘I want to take this opportunity to comment on a technical issue which has become the subject of media discussion … the question of whether a voluntary survey can become a substitute for a mandatory census. It can not’ (Chase & Grant, 2010). Significantly, no country has resorted to a non-mandatory survey only as a substitute for the census.

The combination of a survey with either a register or full census has been a frequent method of overcoming the shortcomings of the survey. Thus basic data are gained from a short census form or the national register, which gains a total number, with basic demographic data, while the detail is filled in with a lengthy survey conducted either on a rotational basis or as a mock census. The United States moved to such a scheme in 2006, with the intention of supplying updated data on a regular basis, thus alleviating the problems of aging census figures (Torrieri, 2007). In view of its negative connotations, the American Community Survey even abandoned the term ‘census’, although a response was legally compulsory. Its intention was to spread the administrative load of completing and processing the former census long-form questionnaire evenly across the decade. Similarly the French introduced a ‘rolling census’ intended to capture the entire population on a five-year cycle (Desplanques, 2008).

Nevertheless, in the 2010 decennial round, some 83% of states which undertook a population count conducted a full mandatory census of the entire population. National consultations for changing the format to surveys or registers are now being contemplated for the 2020 round (United Nations, 2014). In England and Wales, dissatisfaction with small-area census statistics for local government use has led to the examination of alternatives drawn from various administrative registers (Harper & Mayhew, 2012). However, proposals to abandon the compulsory census met strong opposition from the Royal Statistical Society, which described such a move as ‘reckless’ (Royal Statistical Society 2013:2). A prominent academic demographer was more forthright in describing it as ‘stupid’, noting that it was the census which exposed unexpected surprises that were not detected by theoretical projections and administrative data (Dorling, 2013:4). An extensive survey of the subject for the Office for National Statistics concluded that the traditional mandatory lengthy questionnaire-based census still provided the best results, although in 2021 most of the questionnaires would probably be completed online (Skinner et al., 2013). Although generally couched in far milder terms, the conclusion by Statistics Canada contemplating the next census was the same (Statistics Canada, 2014). This was endorsed by the incoming government. Nevertheless, the high costs of the traditional format make the search for alternatives attractive in an era of austerity. Statistics Netherlands was thus able to note that whereas the 2011 census in the United Kingdom had cost US$11.82 per head, the ‘virtual’, but far thinner, census in the Netherlands had cost only US$0.10 per head (Statistics Netherlands, 2014:10).

4. The situation in South Africa

The present state of official statistical collection in South Africa needs to be seen against the background of the apartheid era under the National Party Government. Census and survey results between the 1960s and early 1990s became increasingly problematic as levels of public evasion and opposition increased (Statistics South Africa, 2007a). Doubts were therefore cast upon the accuracy of the statistics (Sadie, 1970). Most significant was the manipulation of the African population numbers, related to the implementation of the Bantustan policies (Lipton, 1972). Further academic articles and reports indicated the degree of scepticism raised with most aspects of South African demographic statistics in the period (Meth, 1988; Phillips et al., 2003; Mare, 2011). Territorial exclusions and substantial levels of undercounting became endemic, as the authors of the final volume of the 1991 census recognised (Central Statistical Services, 1992).

In order to overcome this legacy, the post-apartheid government thus had to improve statistical accuracy and credibility and also encourage popular participation and understanding in the process (Lehohla, 2005). The population censuses were therefore portrayed as nation-building exercises akin to post-colonial models devised by newly independent governments seeking to establish a common identity for people of diverse backgrounds. Extensive publicity campaigns, emphasising their inclusive democratic nature and stressing the benefits to be gained personally and collectively from the data obtained, were conducted under the slogan ‘count us in’ (Statistics South Africa, 1998). In time these developed into an elaborate programme extending from the pre-census consultations to major nationwide education drives at the time of the enumeration, and finally the carefully choreographed result launches, all featuring the Statistician General, Pali Lehohla, in his distinctive yellow suit (Christopher, 2014).

Currently demographic statistics in South Africa are collected primarily by two bodies: the Department of Home Affairs and Statistics South Africa. The Department of Home Affairs is responsible for the registration of births, marriages and deaths, together with recording immigration and emigration, the maintenance of the national population register and the issue of the necessary documentation. Statistics South Africa is responsible for the gathering of information through such means as censuses and surveys of a range of data from population and agriculture to trade and inflation. This is similar in intent to other Commonwealth statistical agencies, such as Statistics Canada, the Australian Bureau of Statistics and the Census of India, in that it is run by professional statisticians rather than politicians and so maintains a degree of independence and distance from the government of the day, to which it is ultimately responsible and on which it is financially dependent. It is notable that the government has supported this degree of independence, stressing the need for sound statistics upon which to base and monitor policies. The former South African minister of finance, Trevor Manuel, exhorted his statisticians to ‘give him the statistics he needs rather than the statistics he likes’ (Kiregyera, 2015:xxii).

South Africa conducts full mandatory census enumerations, with a long-form-style questionnaire, subject to strict confidentiality laws against the disclosure of individual returns. The latest, in 2011, contained 55 personal and 20 householder questions, some of which included up to 14 sub-questions (Statistics South Africa, 2011). The questions had been selected after lengthy public and professional consultation programmes (Statistics South Africa, 2008a). It is notable that the controversial race classification question continues to be asked and meets little resistance in gaining individual identification with the apartheid groupings, because it is viewed as essential to the implementation of government policies of redress. The census cost R3.4 billion, compared with R970 million in 2001, representing a significant increase in costs per head, reaching approximately US$10.00 in 2011 (May & Lehohla, 2005). The 2011 results were processed in the remarkably short period of one year, thus overcoming one of the criticisms of the long-form style of census, namely delays in the availability of data. The 2011 census showed some surprises, when compared with the 2001 census and the projections based upon them, notably in the size of population, due to greater internal and international migration movements and a lowering of mortality rates (Statistics South Africa, 2012). This prompted the Statistician General to defend the principle of census-taking with the comment: ‘The mid-year estimates are estimates based on assumptions and all demographers apply their own assumptions. A definitive arbiter of count is the census and not the model’ (De Wet, 2012:12). Nevertheless, there are (sometimes massive) undercounts which require Statistics South Africa’s own statisticians and demographers to make assumptions to estimate corrections.

Initially the new post-1994 government sought to hold five-yearly censuses, but the projected 2006 enumeration had to be cancelled due to pressure of work in the national statistical office and a mandatory Community Survey or ‘mini-census’ was substituted in its place the following year (Statistics South Africa, 2007b). Because the latter sampled approximately 280 000 households across the country, the volume of work was substantially reduced and the results were published within a few months of collection. Significantly the 2007 survey sought to fill gaps in the statistics for the monitoring of the attainment of the United Nations millennium development goals, which had not been available for consideration for inclusion in the preparations for the 2001 enumeration (Statistics South Africa, 2005). However, small area figures had to be sacrificed in the process and significant problems existed with the data gathered (Statistics South Africa, 2008b). In 2016 the inter-census Community Survey was expanded to sample some 1.3 million households, concentrating on the information needed to monitor the attainment of the United Nations sustainable development goals. Costs preclude the conduct of a quinquennial census, but the annual surveys do not provide the level of data required for the purpose (Jerven, 2014).

In addition, Statistics South Africa in 2002 introduced an annual General Household Survey, succeeding the previous October Household Survey (1994–99) (Aliber, 2009). The annual survey of approximately 29 000 households presents up-to-date figures where both long-term trends and new areas of interest may be discerned. This structured sample compares with the 14.45 million households recorded in the 2011 census. The Survey’s flexibility in filling deficiencies, such as the controversial absence of a religion question in the 2011 census, may be demonstrated by its inclusion in the 2013 General Household Survey (Statistics South Africa, 2014:32). Attempts to compare the various levels of enquiry have revealed some of the strengths and shortcomings of the different methodologies (Weir-Smith, 2014). One of the most significant differences is the availability of the most detailed level of spatially based data, which ranges from small areas, even enumeration tracts, for the census to provinces in the case of the General Household Survey. However, as data are collected by one agency, problems of comparing the definition of terms do not arise frequently, although caution needs to be exercised (Yu, 2009, 2016).

South Africa also possesses a sophisticated, but still incomplete, population register of citizens and permanent residents, which initially had been devised during the apartheid era as a vital element in the enforcement of racially based segregation. The initial register, established in the 1950s, was based on the 1951 census returns, and was concerned with population classification, place of residence and the issue of identity documents (Breckenridge, 2014). In 1991 the racial classification was deleted as part of the democratisation process, but the registration and documentation processes continued. It remains essentially an administrative personal identity register, with little additional information – even current place of residence. The widespread lack of conventional physical addresses to provide the essential spatial framework, estimated at 30% in 2015, remains a significant problem for data analysis (Lehohla & Fleck, 2015). Other national registers are similarly deficient in such coverage. A recent Electoral Court case revealed that only 36% of the 25 million people appearing on the national voters’ role had a full conventional address indicated (Electoral Commission, 2016).

Although national identity numbers, as well as race group, are required when filling in most forms in South Africa, so far there is little acknowledged linkage between different official databases. In particular, the results of census and survey questions on income are not linked to the South African Revenue Services database. A generalised comparison of the 2011 results of the two state databases revealed significant discrepancies, but it appeared that under-recording of income in the census compared with the South African Revenue Service was spread throughout all levels of society (Adcorp, 2013). An earlier, more detailed Canadian comparative study found a more nuanced relationship between the two sets of figures (Brochu et al., 2014). However, there are important gaps in the national register, which currently preclude its use as a survey base list. Undocumented immigration flows into South Africa and to a lesser extent emigration suggest there may be a substantial missing element on the register (Amit & Kriger, 2014). Similarly, homeless people have proved to be remarkably difficult to count and record (Kok et al., 2010). On the contrary, significant advances have been made in the last two decades with the compulsory civil registration of births, which forms an essential part of the creation of a comprehensive national population database, with compliance rising from 24% in 1991 to 95% in 2012 (United Nations Children’s Fund, 2013:33). The introduction of the government National Child Support Grant has proved to be a particularly valuable incentive for the registration of children (Lund, 2012). The provision of disability and old age grants provided a similar encouragement for the registration of other significant sectors of society. A register-based census is thus one of the options which may be available to South Africa at a future date, particularly if the costs of conducting a traditional enumeration become prohibitive (Bah, 2015).

5. Conclusion

South Africa at present conducts full mandatory census enumerations and sample surveys of the population and possesses a basic national population register. It thus has the potential prerequisites for any system of demographic data collection the government might wish to implement, depending upon priorities and costs. In terms of expense, the development of the state population register and even ‘virtual censuses’ has brought significant savings to countries that have implemented them, once the registers have been completed. This suggests the presence of a highly organised bureaucratic state, however, with compulsory registration of addresses and occupations, as well as basic demographic events, linked to other official databases. Such a situation has been essentially contrary to the civic ethos of many Commonwealth countries, which have generally sought to limit government surveillance of the private individual, as illustrated by the current opposition of the Privacy Commissioner of Canada to the linkage of a national population register to the census process as a potential violation of confidentiality undertakings. The South African constitution places a significant emphasis upon upholding human rights. It is supported by powerful independent institutions, including the Office of the Public Protector, charged with the task of holding the government to account for its actions. The severe reservations expressed by the Privacy Commissioner would therefore be relevant to any proposal to link the national population register to the census in South Africa. Yet a lesser degree of state surveillance of the individual in South Africa, as in Canada, does not preclude the current compilation and maintenance of anonymous lists of geographically located physical addresses upon which to base a census or survey, whether including the entire population or only a structured representative sample.

The mandatory census remains the most significant method of enumerating the population and producing a comprehensive set of credible statistics, including the unanticipated surprises. These are often the result of the post-enumeration verification processes, which involve labour-intensive fieldwork. Fortunately, the census in South Africa has not elicited the anti-state or populist opposition rhetoric encouraging non-compliance that governments have faced in some other countries, and participation in the enumeration is still seen positively as an essential part of citizenship in a modern democracy. Furthermore, census statistics can now be produced quickly and the costs of the enumeration have been contained to acceptable levels. These are all positive assets for the continuation of probably the most satisfactory means of obtaining reliable comprehensive demographic and other information on the population of the country – the mandatory census. It is therefore suggested that the current debate on the future of the census is primarily a warning presented by the Canadian example of making precipitous changes for purely party political reasons, which may lead to significant gaps emerging in the national sequence of statistics and doubts as to their integrity. There is also the danger that the official statistics which are produced may lose their credibility by becoming popularly perceived as the statistics which the political party in government likes rather than those it needs. However, at present the restoration of statistical integrity and independence in South Africa since the demise of apartheid appears to be guarded carefully. Nevertheless, public trust in the credibility of official census statistics has to be renewed constantly.

Acknowledgements

The financial assistance of the National Research Foundation towards the costs of this research is hereby acknowledged. The opinions expressed in this article and the conclusions arrived at are those of the author and are not necessarily to be attributed to the National Research Foundation.

Additional information

Funding

This work was supported by the National Research Foundation [grant number 80998].