Age reporting by and for older people in Uganda: relationships with frailty, human capital and population registration

Authors:
Alice Reid University of Cambridge, UK

Search for other papers by Alice Reid in
Current site
Google Scholar
Close
,
Allen Kabagenyi University of Makerere, Uganda

Search for other papers by Allen Kabagenyi in
Current site
Google Scholar
Close
,
Stephen Ojiambo Wandera University of Makerere, Uganda

Search for other papers by Stephen Ojiambo Wandera in
Current site
Google Scholar
Close
,
Joshua Beinomugisha University of Makerere, Uganda

Search for other papers by Joshua Beinomugisha in
Current site
Google Scholar
Close
, and
Sarah Walters London School of Hygiene and Tropical Medicine, UK

Search for other papers by Sarah Walters in
Current site
Google Scholar
Close
Full Access
Get eTOC alerts
Rights and permissions Cite this article

In this article, we investigate the accuracy of age reporting by people aged 60 and older and proxy reporting by their carers in a peri-urban area of Uganda, and analyse the factors that influence reporting by both groups. We find a high level of age heaping on terminal digits 0 and 5, indicating poor knowledge of age. Contrary to other studies, we find that literate people were more likely to exhibit age heaping. We link this to the absence of birth registration for this cohort and the introduction of National Identification (ID) cards in Uganda five years before our survey. We conclude that age heaping is better interpreted as an indicator of registration machinery than of human capital. We also find that the health, functional capabilities and education of older people influenced the age ascribed to them by their carers. Carers who knew the older people less well were more likely to overestimate their age, and carers of healthy and more educated people were more likely to report a younger age than that reported by the older people themselves. Where people don’t know their age, the age they report may also be influenced by their health and capabilities, making it difficult to establish true relationships between chronological age and outcomes such as health. In many disciplines, self-reported age or age reported by proxy respondents is accepted uncritically by researchers, but our study shows that in peri-urban Uganda age reporting remains approximate and biased, and this has strong implications for appropriate targeting and monitoring of interventions to support healthy ageing in such contexts.

Abstract

In this article, we investigate the accuracy of age reporting by people aged 60 and older and proxy reporting by their carers in a peri-urban area of Uganda, and analyse the factors that influence reporting by both groups. We find a high level of age heaping on terminal digits 0 and 5, indicating poor knowledge of age. Contrary to other studies, we find that literate people were more likely to exhibit age heaping. We link this to the absence of birth registration for this cohort and the introduction of National Identification (ID) cards in Uganda five years before our survey. We conclude that age heaping is better interpreted as an indicator of registration machinery than of human capital. We also find that the health, functional capabilities and education of older people influenced the age ascribed to them by their carers. Carers who knew the older people less well were more likely to overestimate their age, and carers of healthy and more educated people were more likely to report a younger age than that reported by the older people themselves. Where people don’t know their age, the age they report may also be influenced by their health and capabilities, making it difficult to establish true relationships between chronological age and outcomes such as health. In many disciplines, self-reported age or age reported by proxy respondents is accepted uncritically by researchers, but our study shows that in peri-urban Uganda age reporting remains approximate and biased, and this has strong implications for appropriate targeting and monitoring of interventions to support healthy ageing in such contexts.

Introduction

Accurate age reporting is fundamental for research on ageing and the definition of older people, which is often based on a particular age as a cut-off point. Even studies which seek to identify the way that experiences differ as people grow older – including medical and social gerontological studies of frailty, mortality, social needs and relationships – depend on accurate information on chronological age as a point of comparison. And research on more subjective and personal experiences of the ageing process may contrast people’s subjective or felt age with their chronological age, thus relying on accurate assessment of the latter.

Such studies rely on the age reported by or for older people, but researchers rarely assess whether those ages are accurate. In many circumstances, it is reasonable to assume that they are: in high-income countries, people need to report their age regularly and they possess documents which record date of birth, which they are able to interpret because of high levels of literacy and numeracy. In such settings, people may need to give their age often and, as a result, do not need to refer to a document or do a calculation to find their age. Where people do not need to report their age frequently, where they do not have an accurate record of their date of birth, or where literacy or numeracy is poor, people’s grasp of their age may be weaker, manifesting in age misreporting. Random errors in reported age tend to cancel out, but systematic errors are common in certain circumstances and can seriously bias results. Age reporting is particularly problematic among older people in Africa (Aboderin, 2010), where ‘data [on the age of older people] available are so problematic that any conclusions about age-related health and welfare and their evolution over time and space are potentially compromised’ (Randall and Coast, 2016: 143–4). This article examines age reporting by and for older people in a peri-urban community in Uganda and discusses influences on the way that age is reported in this setting.

The systematic misreporting of age takes two main forms: age heaping and age shifting (Johnson et al, 2022: 13–4). Age heaping, or digit preference, is the tendency to report age in round numbers, often numbers ending in 0 or 5. A variety of indexes, such as Whipple’s index, quantify the degree of age heaping. Age shifting is the systematic over- or underestimation of age, and it is frequently concentrated in particular age-sex groups. For example, young women are more likely to overstate their age, but middle-aged women have historically been more likely to understate it (Ewbank, 1981: 60-3; Census of England and Wales, 1881: 18). This may be linked to a desire to be on one side of a legal age barrier (such as being over the minimum age for renting a property) or to avoid a lengthy section of a questionnaire (for example, in the Demographic and Health Survey – the DHS Program – women under the age of 50 have many more questions to answer). Age heaping is much easier to detect in data than age shifting, and so analyses of the quality of age reporting tend to concentrate more on the former.

Age heaping tends to increase with age, and there is some evidence that over a certain age, overestimation of age also increases with age (Coale and Kisker, 1986; Smit et al, 1997). Age misreporting may produce misleading mortality rates (Coale and Kisker, 1986; Coale and Li, 1991; Elo and Preston, 1994) and can lead to distorted dependency ratios (the proportion of children and/or older people in a population) (Randall and Coast, 2016) and population projections (estimations of what the population will look like in the future) (West et al, 2005). If reported age reflects the health status of older people, it becomes impossible to assess whether people are getting healthier as they get older or whether additional years of life are spent in ill health. Without reliable knowledge of chronological age, it becomes impossible to assess any discrepancy with subjective or felt age. Therefore, although individuals themselves may not place a high value on knowing their precise chronological age, accurate age reporting is still important for generating reliable demographic rates, for planning services based on needs and for understanding the ageing process.

This article is derived from a project entitled ‘Pictures of ageing in Uganda – a partnership to explore demographics, phenotype and self-perception in a community of older people’. The project brought together academics from medicine, psychology, psychiatry, demography, qualitative social science and art. The team worked with peri-urban communities near Kampala, Uganda, to co-develop an interdisciplinary pilot study in the area of health and ageing. Due to the importance of accurate age reporting for understanding experiences of ageing, there was a strong focus on the reporting of age, and this article reports the findings from this component of the research. We examine age heaping by the older people who participated in the research (referred to as ‘respondents’) and their carer, companion or neighbour (referred to as ‘carers’ for the sake of brevity), who provided proxy reports of the older people’s age. We also look at respondents’ consistency of age reporting, and we examine age shifting as reported by the carers.

Literature review

The literature on age heaping is concentrated in two academic fields: demography and economic history. The former focuses on age heaping as a marker of the quality of demographic data, and the latter focuses on age heaping as an indicator of human capital. Despite this difference in focus, these fields use the same tools and share an interest in calculating age heaping for different groups in society, and there is also some overlap in the places and times which have been studied. Generally, age heaping is more common among older people, people with little or no education, rural dwellers and women, as observed in both the economic history (A’Hearn et al, 2009; Crayen and Baten, 2010; Földvári et al, 2012; Tollnek and Baten, 2014) and demography (Fayehun et al, 2020) literature. Some studies found that married people, particularly married women, reported age more accurately, and this is attributed to the woman pegging their own age to that of their husband rather than to independently better reporting (Földvári et al, 2012). However, Elo et al (1996) found that women were more likely than men to report their spouse’s age at death accurately, and Malik (2021) found that in India women reported age more accurately than men. In some circumstances, religion has also been associated with differences in age heaping, with Jews exhibiting less age heaping than Catholics (Tollnek and Baten, 2014; Juif et al, 2020).

The correlation between age heaping and education or literacy and numeracy has often been interpreted as a causal link, the implication being that people with little or no education are unable to keep track of, or calculate, their age. For example, with reference to recent censuses in different African countries, Mba writes that a ‘high rate of illiteracy, especially among elderly people [sic], … is responsible for their inability to keep accurate records of their dates of birth’ (2014: 25; see also Lyons-Amos and Stones, 2017; Francis et al, 2019; Shipanga and Shinyemba, 2023). In the economic history literature, this relationship has led to age heaping becoming one of the most frequently used measures of numeracy and, therefore, of human capital (A’Hearn et al, 2009; Crayen and Baten, 2010). Many economic historians apply a literal interpretation of this connection, seeing age heaping as a direct indicator of mathematical skills or cognitive ability (Tollnek and Baten, 2017; Perrin, 2020). However, others argue it is also a function of census-taking practices (Elo and Preston, 1994; Elo et al, 1996; Spennemann, 2017), and Szołtysek et al (2018) found that more careful questions about age and better training of enumerators improve the quality of age reporting.

In general, the reporting of age is more accurate when information is gathered on date of birth rather than age last birthday (West et al, 2005), though using date of birth can also lead to heaping on digits other than 0 or 5 due to preferences for rounded years of birth.1 Where there is official registration and record keeping, this can improve age reporting in censuses and surveys; for example Rosenzweig (2021) found that people with birth certificates were less likely to report heaped ages. Increasingly, scholars are recognising that literacy and age reporting are highly correlated because they have common roots in a process of administrative and cultural modernisation (Spennemann, 2017; A’Hearn et al, 2022b). Age heaping is likely to be more common where people do not need to continually rehearse their age for administrative, legal or social reasons, where they do not have documentation proving their age or date of birth, and where there is little cultural importance placed on knowledge of age or birthdays. Age heaping may, therefore, be a consequence of lack of knowledge of age, but the presence or absence of registration and the ways that censuses and surveys are carried out (including proxy reporting – that is, people reporting on behalf of others) are also likely to be influential.

Proxy reporting of age (which takes place in many household censuses and surveys, and is always the mode of reporting when deaths are recorded) is likely to be less reliable than individuals reporting their own age. Moreover, West et al (2005) found that in the United States Census 2000, proxy reporting of age by non-household members was significantly more prone to age heaping than proxy reporting by family members. Elo et al (1996) found that spouses reported age at death more accurately than other proxy informants and that wives were more accurate than husbands. Lankoandé et al (2022) found that in Burkina Faso the quality of reporting age at death was much worse than the quality of reporting age of living people, and the same pattern was found for the United States (Elo et al, 1996; Preston et al, 1996).

The ability to perceive age in the physical features of another may influence proxy reporting. Studies of age perception based on facial photography indicate that it is more difficult to estimate the age of older people than the age of younger people, and that people also tend to be better at estimating the age of people in their own age group (Voelkle et al, 2012). This is particularly important in censuses and surveys given that data for the whole household may be provided by a single respondent. It has also been observed that in some studies, assessments of age are informed, or even decided, by the interviewer (Pardeshi, 2010; Randall and Coast, 2016).

Estimated age is also likely to be influenced by cultural understandings of ageing – who is considered to be ‘old’ – which depend on, for instance, a person’s health, social role and social status (Hausknecht et al, 2020). Thus, individuals who appear to conform to markers of old age may be more likely to be assigned an older age by a proxy respondent. For example, ethnographic studies such as that by Glascock and Feinman (cited in Kowal and Dowd, 2001) suggest that capabilities (related to, for example, invalid status, senility or physical characteristics) can affect others’ perceptions of whether a person is old, and that social role (related to work patterns, the adult status of children, the experience of menopause and so on) may also be influential. In many African societies, high social status confers seniority, and thus individuals with higher social status may be assigned a higher age (Sagner et al, 2002). In most societies, frailty and declining physical functionality are seen as markers of old age, and therefore older people who appear more frail and have more limited functionality are considered older. This might also feed into people’s perception of their age. Studies of age identity (subjective, perceived or felt age) in the fields of psychological ageing and social gerontology have found that it is common to feel younger than one’s chronological age (particularly among older people), but that the experience of illness makes people feel older (Morelock et al, 2017; Bordone et al, 2020; Pinquart and Wahl, 2021; Demir Erbil and Hazer, 2022). Biomedical studies have found that perceived age is generally a good indicator of biological age (which is related to the condition of the cardiovascular, metabolic or immune systems, and may differ from chronological age; Christensen et al, 2009; Jones et al, 2019), possibly because perceived age is heavily dependent on health, capabilities and even facial expression; Voelkle et al (2012) found that the age of happy faces tends to be underestimated.

Age reporting in Africa

Age reporting in Africa is generally acknowledged to be poor quality. Some overview studies on human capital have included African data; for example, Crayen and Baten (2010) found high levels of age heaping for those born in the 1940s and even higher levels for those born in the 1890s, but it is unclear what countries were used in the analysis for each of these periods. Research specifically on Africa has found decreases in age heaping over the long term from 1730 to 1970 (Cappelli and Baten, 2021), but continued high levels and little consistency in recent trends. For example, Mba (2014) found that age reporting in African censuses has generally improved over time, but that there are still high levels of inaccuracy across the continent, with better reporting in countries in Southern Africa than those in West Africa, and East African countries falling between those. In contrast, evidence from the DHS indicates little change over recent decades (Lyons-Amos and Stones, 2017).

In terms of which groups are more likely to exhibit age heaping, Fayehun et al (2020) found that in the Nigerian DHS, age heaping was more prominent among men, people with little or no education and rural dwellers. Capelli and Baten (2021) found that former British colonies had lower levels of age heaping, which they attributed to differences in colonial education systems. Economic historians keen to substantiate a link between mathematical ability and age heaping in present-day Africa found that age heaping was greater among the parents of children who performed less well in mathematics tests (Baten et al, 2022). However, the evidence presented by other studies is mixed. For example Francis et al (2019) found that in Ghana, patients with medical insurance exhibited more age heaping than those without insurance; they attributed this to greater levels of illiteracy among those who were insured, but did not explain why this might be so.

However, few African studies have looked specifically at the accuracy of age reporting by or for older people and, consequently, relatively little is known about age reporting among older people in Africa. A project to establish ‘a minimum data set on ageing and older persons in sub-Saharan Africa’ was initiated in the early 2000s (Ferreira and Kowal, 2013), but this focused on identifying the sort of data that would be useful and did not specifically consider the quality of such data (Randall and Coast, 2016). Lankoandé et al (2022) and Wak et al (2017) compared age reporting in censuses and established demographic surveillance sites (DSSs) in Burkina Faso and Ghana, respectively.2 They found considerably more age heaping in census data than in DSS data, particularly among older people, which they attributed to greater rigour and consistency in the DSS data collection process in Burkina Faso (Lankoandé et al, 2022) and more date of birth documentation for younger people in Ghana (Wak et al, 2017). The most comprehensive analysis of age reporting by older people in Africa is provided by Randall and Coast (2016), who compared the quality of age reporting among older people in censuses and the DHS in different countries. They report that while, in some African countries, knowledge of age or date of birth is improving among younger generations due to increased schooling and more administrative demands for date of birth information, there are still considerable problems of age reporting among older people and those who have little or no education. They argue that age reporting among older people is particularly inaccurate in sub-Saharan Africa because of the widespread social irrelevance of knowing absolute age, although relative age is important.

In the African context, there are few studies comparing older people’s chronological age with the age other people consider them to be. However, researchers have compared subjective and chronological age among older Africans in Senegal (Macia et al, 2012; Macia et al, 2019) and in Burkina Faso (Schönstein et al, 2021). While this work confirms the finding from the wider literature that people’s age identity is linked to their health, it also finds considerably less discrepancy between individuals’ felt age and their chronological age than is the case in higher-income countries. For example, in the study in Burkina Faso, self-perceived age barely differed from chronological age; the authors suggest this might be because youthfulness has a lower value in this setting than in high-income contexts (Schönstein et al, 2021). Similarly, felt age and chronological age were identical for 76 per cent of the sample in Senegal; the authors argue that compared to older people in higher-income countries, Senegalese older adults may find it harder to ‘ignore their corporeality’ (Macia et al, 2019: 829). We interpret this to mean that with a relative lack of products and services which can make life easier, older people in Senegal experience more age-related limitations to daily life. While these observations may well have merit, it is notable that these studies did not consider the accuracy of the older people’s reports of their chronological age. A lack of knowledge of actual age or birth date might mean that their reports of chronological and felt age were similar not because both reflected chronological age, but because both reflected felt age.

The literature therefore suggests that age reporting by older people in an African context is likely to be poor, and that this might be linked to low numeracy or mathematical ability or to lack of knowledge of actual birth date due to relatively low administrative demands for birth date reporting. It also suggests that older people’s reporting of their age, and the age that other people consider them to be, might depend on the older people’s health status, capabilities and appearance. We consider these aspects in our analysis of age reporting by and for older people in a peri-urban area of Uganda.

Data and methods

The ‘Pictures of ageing’ project developed a pilot study of 150 people thought to be aged 60 and older in three villages in the Busukuma subcounty in Wakiso district, about 25 kilometres from Kampala. A list of older people in these villages was provided by village officials, and the individuals listed (the ‘respondents’) were visited by the project team, with a family member or other companion (the ‘carers’) present.

A questionnaire was carefully constructed so that, first, the carer was asked what age they thought the respondent was. Next, the respondent was asked to give their own age. Then, they were asked more detailed questions about their date of birth, whether they had any documentation showing their age or date of birth, and their memory of a series of historical events (to determine which major events they remembered), and this was followed by a series of socioeconomic, demographic, household and health-related questions.3 The research took place in a locality with a Christian tradition (although 17 per cent of our sample were not Christian), and for those who reported they had been baptised in the vicinity, we searched for baptism records at their church in order to obtain a date of birth reported at baptism. Baptism often occurs in infancy or childhood, so the birth dates reported then may be more accurate than those reported later in life (Helleringer et al, 2019).

We use a range of analyses to consider reporting of age in our study population: analysis of age heaping by respondents and carers; analysis of the consistency of respondents’ age and date of birth reporting; and analysis of the discrepancy between age reported by respondents and carers. These approaches are described next.

Age heaping

Analysis of age heaping is possible through a range of indices which detect heaping on 0, 5 or other digits – for example, Bachi’s index, Myers’ blended index, the UN age-sex accuracy index and Whipple’s index. The Myers’ blended index can identify heaping on a variety of different terminal digits, as can some adaptations of Whipple’s index (Shryock and Siegel, 1976; Spoorenberg, 2007; A’Hearn et al, 2009). However, the most widely used and the easiest to compute and interpret is the original Whipple’s index, which is very suitable for places that follow the most common pattern of heaping; that is, on terminal digits 0 and 5. Previous studies confirm this pattern for Uganda (see Mba, 2014), and we chose to use Whipple’s index in our analysis.

Whipple’s index is calculated by summing the number of persons aged 23−62 years (inclusive) who report ages ending in 0 or 5, dividing that sum by the total population aged 23−62, and multiplying by five. This produces an index which is related to the quality of the age data. The index can range between 100 (no digit preference in age reporting) and 500 (everyone reports an age ending in either 0 or 5). An index of 100 to 104 is interpreted to mean ‘highly accurate’ data, 105 to 109 indicates ‘fairly accurate’ data, 110 to 124 indicates ‘approximate’ data, 125 to 174 indicates ‘rough’ data, and 175 and over indicates ‘very rough’ data.

Whipple’s index is generally calculated for ages from 23 to 62, and sometimes up to 72. Older ages are omitted because the index assumes rectangularity; that is, the same number of people in each single year age group across a ten-year age range. This would not be the case for ages over 72, as high mortality rates mean older cohorts become smaller (Crayen and Baten, 2010); thus, applying Whipple’s index to older ages would overestimate age heaping. Nevertheless, Randall and Coast’s (2016) modified Whipple’s index (Whipple60) for older people simply sums the people aged 60–94 whose ages end in 0 or 5 and divides by the total number aged 60 and over. They use 94 as the upper limit since some sources report ages higher than this in a single age group (95 and over) and because they find good agreement between the Whipple and Whipple60 indices. However, as mentioned, the Whipple60 will produce inflated estimates of age heaping because of the non-rectangularity of older ages, and this is exacerbated when the starting age for the calculation is age 60 rather than age 58.4

We use Randall and Coast’s Whipple60 to allow comparison with their study. To address the non-rectangularity issue, our statistical tests (chi-square) examine whether the number of ages ending in 0 or 5 is different to the number expected according to a linear decline in ‘true’ cohort size with increasing age, starting at age 60 and declining to zero people at age 100. Under this scenario, we would expect 22.1 per cent of the population to ‘truly’ have a terminal digit of 0 or 5, and 12.3 per cent to have a terminal digit of 0.

We report the Whipple60 index by a series of variables – sex, precision of data of birth provided, reading ability, writing ability, education, marital status and highest occupation – and we present the results of a logistic regression on the likelihood of reporting an age ending in 0 or 5 to identify the most important correlates of age heaping.

Precision and consistency of age reporting by respondents

We gathered various pieces of information from respondents about their age: all gave their age last birthday and their date of birth.5 People who had been baptised reported the date of their baptism and whether they had a baptism certificate. If a participant offered a day, a month and a year for the dates, these were recorded. If a respondent only offered a birth year (without specifying the day and month), their birth date was recorded as 1 January (or 1/1) that year. If they said that they were born in a specific month of their year of birth, but did not give a day, their birth date was recorded as the first of that month (for example, 1 August or 1/8). We classified these reported dates into three levels of precision: high precision (day ≠ 1); medium precision (day = 1, month ≠ 1); and low precision (day = 1, month = 1).6

We also considered consistency of reporting, allowing two different ways of returning a ‘consistent’ age: (1) if reported age was the same as that calculated using the date of the interview and the reported date of birth; (2) if reported age was the same as the difference between the year of birth and the year of interview. For respondents whose birthday came before the date of interview (including all those with low precision), these two calculations were the same. For respondents whose reported birthday came after the date of interview, these calculations always differed by one year, but we allowed a consistent answer for either to count as consistent.

Discrepancy in age reported by respondents and carers

In order to investigate whether the age that carers assign to respondents is affected by the frailty or functional capacity of the older person, or by the carers’ relationship to the respondents, we analysed the difference between respondents’ self-reported age and their age as reported by carers. After carrying out descriptive analysis, we performed two logistic regressions to analyse: (1) the likelihood that carers reported an age at least five years younger than the respondents’ self-reported age; and (2) the likelihood that the carers reported an age at least five years older than the respondents’ self-reported age.

Results

Age heaping by respondents and carers

Figure 1 shows the distribution of ages returned by respondents and carers. Despite the low number of cases, there is strong age heaping on ages ending in 0 among reports from both groups, particularly carers. There is also some evidence of heaping on ages ending in 5, though this is less pronounced than for ages ending in 0.

Column graph showing the number of people reported by a) respondent and b) carer to be each exact age. More people than expected are reported to be ages ended in 5 and particularly 0, especially for ages reported by carers.
Figure 1:

Distribution of ages reported by respondents and carers

Citation: Journal of Global Ageing 1, 1; 10.1332/29767202Y2023D000000003

Table 1 shows the Whipple60 index and whether the age reported is significantly different to that expected if there were no age heaping. The higher the index, the less accurate the reporting of age. There is remarkable consistency in the table, with only one value (low precision of date of birth provided) which does not indicate rough or very rough data, and even this value indicates that the data is only approximate.

Table 1:

Whipple60 index of age heaping

Variable Whipple60a Significance

(p-value)b
Number
All respondents 167 0.00 150
All carers 253 0.00 150
Sex
 Men 140 0.35 43
 Women 177 0.00 107
Precision of date of birth provided
 High 202 0.00 62
 Medium 156 0.08 24
 Low 119 0.85 21
Ability to read
 Cannot read 148 0.18 54
 Can read 177 0.00 96
Ability to write
 Cannot write 125 0.61 52
 Can write 189 0.00 98
Level of education
 None or incomplete primary 158 0.02 98
 Primary or incomplete secondary 159 0.14 41
 Secondary or higher 273 0.01 11
Marital statusc
 Married/cohabiting 183 0.01 52
 Widowed 147 0.18 58
 Divorced/separated 167 0.09 39
Highest occupation
 Shopkeeper and higher 167 0.18 24
 Skilled/semi-skilled labourer 200 0.03 25
 Unskilled labourer 125 0.71 28
 Agricultural worker 174 0.04 43
 Missing 167 0.14 30

Notes: All calculations refer to responses by respondents only, except that for ‘All carers’ which refers to responses by carers only.

a A Whipple60 index of 100 indicates that the expected number of people return an age ending in 0 or 5. The higher the Whipple60 index, the higher the likelihood of reporting an age ending in 0 or 5. b The p-value indicates whether the number returning their age with a terminal digit of 0 or 5 is significantly different from the expected number (22.1 per cent). Results which are statistically significant at the 10 per cent level are shown in bold. c One person in the sample had never married.

As suggested by Figure 1, the age heaping index is higher for ages reported by carers. The index for all respondents is 167 (rough) and the index for all carers is 253 (very rough). For both respondents and carers, the number returning their age with a terminal digit of 0 or 5 is significantly different from the expected number (22.1 per cent of the sample). Not all values are statistically significant, probably due to the small number of respondents.

As expected, women exhibit more age heaping than men. Other variables, however, do not show expected patterns. Contrary to other studies, people who are married or cohabiting exhibit more heaping than other marital status groups, and none of the indicators which might be interpreted by economic historians as representations of human capital show expected results: those who can read, can write, have the highest levels of education and report date of birth with a high level of precision are all more likely to report an age ending in 0 or 5. This provides little support that age heaping can be interpreted as an indicator of numeracy, or indeed of human capital of any sort. Occupation does not show a consistent pattern; nor does religion (not shown).

Many of the factors investigated are likely to be highly correlated. To ascertain which were the most important determinants of age heaping by respondents, a multivariate logistic regression was performed with those reporting an age ending in 0 or 5 (Table 2; for descriptive statistics, see Supplementary materials).

Table 2:

Multivariate logistic regression of age heaping by respondents (reporting an age ending in 0 or 5)

Variable Crude ORa 95% CI Adjusted ORb 95% CI
Sex
 Male 1.0 1.0
 Female 1.4 0.66–3.09 1.8 0.79–4.09
Ability to write
 Cannot write 1.0 1.0
 Can write 1.8 0.86–3.85 2.1* 0.94–4.70
Self-rated overall health statusc
 Very bad/bad 1.0 1.0
 Moderate/good 2.7** 1.20–5.89 2.6** 1.18–5.90
 Very good 1.5 0.39–5.68 1.3 0.33–5.12
N 150
Pseudo R2 0.06

Notes: OR = odds ratio; CI = confidence interval.

a Crude OR shows results for the variable without other controls. b Adjusted OR shows results for the full model. c Health ratings cover the last 30 days. * p < 0.1; ** p < 0.05; *** p < 0.01. Results which are statistically significant at the 10 per cent level are shown in bold.

Only one variable had a category with a statistically significant relationship with age heaping in the crude models, although the magnitude of results matched the results of the Whipple60 analysis. Those who rated their overall health in the last 30 days as moderate or good (as opposed to very bad or bad) exhibited significantly more age heaping. Those who rated their health as very good showed no statistically significant difference in age heaping compared to those who rated their health as very bad or bad.

The full model included only variables for sex, ability to write and self-rated overall health status. Other variables were omitted due to collinearity or lack of significance and the interests of parsimony. In this model, after controlling for sex and self-rated health status, respondents who could write were over twice (adjusted odds ratio = 2.1; p = 0.07) as likely to report an age ending in 0 or 5 as those who could not write.

Precision and consistency of age reporting by respondents

To further examine older people’s knowledge of their age, we considered whether the age reported by respondents matched their age calculated by taking the difference between their reported date of birth and the date of the interview. Arguably, reporting an age which is consistent with recorded date of birth is a better measure of mathematical ability or numeracy than reporting an age ending in 0 or 5, and Table 3 demonstrates that, contrary to our findings on age heaping, this measure shows an expected relationship with ability to read and ability to write, and with education level. However, there is very little relationship between consistency in age reporting and the reporting of an age ending in 0 or 5. It seems that among the older people in our study, numeracy or human capital are not well represented by age heaping.

Table 3:

Consistency between respondents’ age reporting and age calculated using date of birth

Number Consistent (%) Inconsistent (%)
All 147 84.4 15.6
Precision of date of birth provided
 High 62 95.2 4.8
 Medium 21 85.7 14.3
 Low 64 73.4 26.6
Ability to read
 Cannot read 54 77.8 22.2
 Can read 93 88.2 11.8
Ability to write
 Cannot write 52 75.0 25.0
 Can write 95 89.5 10.5
Level of education
 None or incomplete primary 98 78.6 21.4
 Primary or incomplete secondary 39 94.9 5.1
 Secondary or higher 10 100.0 0.0
Age heaping
 Ends in 0 or 5 50 82.0 18.0
 Ends in other digit 97 85.6 14.4

Table 3 also shows that people who reported their date of birth with high precision were more likely to report a consistent age. This, together with the finding (shown in Figure 2) that those who are able to write were considerably more likely to report their date of birth with precision, represents a conundrum: as might be expected, those with higher levels of education and higher literacy were more likely to give a precise date of birth, but they were also more likely to report an age ending in 0 or 5. The implication of this is the unlikely scenario that each five or ten years there was a cohort born who achieved higher literacy.

Stacked column graph showing the percentage of respondents reporting their date of birth with high, medium and low precision, for those who cannot and can write separately. Among those who cannot write, nearly 70 percent report date of birth with low precision, and less than 20 percent with high precision. Among those who can write around 30 percent report date of birth with low precision, and over 50 percent with high precision.
Figure 2:

Precision of date of birth reported by respondents, by ability to write

Citation: Journal of Global Ageing 1, 1; 10.1332/29767202Y2023D000000003

The ways in which people might learn or rehearse their birth dates offer a potential reason for this unexpected finding. We asked respondents about date of birth documentation, and many reported that they had a baptism certificate or National ID card. For those baptised as infants, a baptism certificate might give an accurate date of birth. The majority of participants (129) were baptised and 116 gave a date for their baptism, 38 of which indicated a baptism during infancy. Of those respondents who had infant baptisms, 16 reported having a baptism certificate.7 It therefore seems unlikely that many of the individuals in this study had documentation from the time of their birth with which to validate their birth date. Nevertheless, there remains potential for use of baptism records to validate age reporting for older people baptised in infancy as long as careful consideration is given to selection of study site (HelpAge International, 2011).8

At first, National ID certificates also appear unhelpful for accurate age reporting, as the ID system was introduced in 2014 (Resilient Africa Network, 2019), just five years before our survey. If people had no documentation with date of birth at the time of registration for the National ID card, their identity could be verified by the Parish Citizenship Verification Committee (Handforth and Wilson, 2019). In such cases, a date of birth to go on the card would have been chosen or assigned, and it is possible that dates of birth yielding a round age in 2014 were more likely to have been selected. Those dates of birth will also have been round ages in 2019, at the time of the survey.

Of the 150 respondents, 118 had an ID card. Although those who could write were no more likely to have a card than those who could not write (78 per cent and 81 per cent, respectively), 49 per cent of those with ID cards reported their birth date with high precision as opposed to 17 per cent of those without a card. Also, 14 per cent of those with an ID card reported an age which was inconsistent with their date of birth compared to 23 per cent of those without a card. It is therefore possible that provision of ID cards ‘fixed’ a date of birth for many people who previously did not know theirs, and in doing so it ‘baked in’ a certain degree of age heaping. This might explain the finding that more literate people in this survey are more likely to report a rounded age through one of two mechanisms: first, more educated people were quicker to sign up for ID cards (Resilient Africa Network, 2019: 21; van der Straaten, 2022) and will therefore have been more likely than those with less education to have received a card in 2014; and, second, more literate people may be more likely to consult or refer to their ID card and therefore to have memorised their assigned birth date.

Discrepancies in age reported by respondents and carers

Next we consider discrepancies in the reporting of respondents’ age by the respondents and the carers. Our findings based on the Whipple60 index (see Table 1) indicate that the proxy reports of respondents’ age, by carers, were less precise than the ages reported by respondents themselves. This is expected, but it remains interesting to investigate whether there are systematic patterns in the imprecision of age reporting, such as whether carers were more likely to report an age older or younger than the respondents themselves. If there are systematic patterns, do these reflect characteristics of the proxy or of the older person?

It is important to remember that respondents may not know their age any more accurately than carers. A finding that, for example, carers were more likely to give an older age could be due either to carers overestimating the older person’s age or to older people underestimating their own age.

Figure 3 shows discrepancies in age as reported by respondents and carers, broken down by a variety of variables. In the graph, the left side shows the percentage of cases where the carer gave an age lower than the age the respondent gave – dark grey is used for ages five or more years lower and mid grey is used for ages one to four years lower. The right side of the graph shows the percentage of cases where the carer gave an age higher than that given by the respondent – dark grey is used for ages five or more years higher and mid grey is used for ages one to four years higher. The lightest shade, in the middle, indicates that carer and respondent gave the same age for the respondent.

Stacked bar chart showing the percentages of respondents where the carer-reported age is higher, the same and lower than the respondent-reported age. Different bars are shown for men and women, and different categories of physical activity, difficulty carrying out day to day activities, self-rated health =, and carer-respondent relationship. The biggest differences are for the health and activity categories, where better health and greater activity are correlated with carer stating age to be younger than the respondent-stated age.
Figure 3:

Discrepancies in age reported by respondents and carers, by various indicators

Citation: Journal of Global Ageing 1, 1; 10.1332/29767202Y2023D000000003

In age reporting for male respondents, roughly the same proportion of carers reported a younger age and an older age than the age reported by the respondent. However, when the respondent was a woman, carers were more likely to give a lower age than that given by the respondent. This could be due to carer characteristics (because women tend to be younger than their husbands and also to live longer, their carers are less likely to be a spouse who might know their age better) or respondent characteristics (women tend to live longer and spend more years in poor health, so carers may judge them to be older).

Turning to the relationship between carers and respondents, close relatives, such as spouses and children, were more likely than other carers to report the same age as the respondent. The graph also indicates that they were also more likely to give a younger than an older age, which could be due to either under-statement by carers or over-statement by respondents, although the former is more likely as it is not clear why respondent accompanied by a spouse or child might be more prone to exaggeration than others. Grandchildren, however, were more likely to report an older age than the respondent. This might reflect low ability of young people to gauge age and the tendency for children to think that older people must be very old. However, other studies investigating perceived age suggest that age and sex of the person guessing the age do not usually affect the accuracy of the result (Jones et al, 2019). Neighbours were as likely to report an age that was older as an age that was younger, and ‘other’ carers (predominantly friends and maids), who perhaps knew the respondents least well, were least likely to report the same age as the respondents reported themselves.

In relation to the health of the respondent, carers of people in good health were most likely to offer an age five or more years younger age than the respondent provided. This held for various different measures of health which carers might notice, including respondents’ self-rated health, difficulty carrying out day-to-day activities and degree of physical activity. The carers of people in poor health were more likely to return an older age for the more ‘objective’ measures of ability to carry out day-to-day activities and degree of physical activity.

The relationship between carers and respondents might be related to respondent health. For example, older people, who may have been in poorer physical health, were less likely to have a living spouse to look after them. Multivariate logistic regression was therefore performed to tease apart the effects of different influences.

Table 4 shows crude and adjusted odds ratios for the carer reporting an age five or more years lower or, in a separate regression, five or more years higher than that reported by the respondent. The crude results mirror those in Figure 3. Compared to spouses, all other carers were more likely to report an age five or more years higher than that reported by the respondent, and this was particularly likely where the carer was not a relative or neighbour. Those older people who were more physically active were unlikely to be allocated much older ages by their carers (difficulty carrying out day-to-day activities gave a similar result but was not included in the adjusted model due to collinearity). Interestingly, the carers of people who can write were considerably less likely to suggest an age five or more years higher than an age five or more years lower (again, a similar result was obtained for ability to read, but this was omitted from the adjusted model due to collinearity). It is possible that those who can read and write appear more mentally agile to others than those who cannot, leading to a lower estimate of age.

Table 4:

Multivariate logistic regression of the carer reporting an age five or more years lower or higher than the age reported by the respondent

Variable Carer reports age five or more years lower

Carer reports age five or more years higher

Crude ORa 95% CI Adjusted ORb 95% CI Crude ORa 95% CI Adjusted ORb 95% CI
Age of respondent 1.0 0.96–1.08 1.1 0.98–1.13 1.0 0.93–1.05 0.9 0.87–1.01
Sex
 Male 1.0 1.0 1.0 1.0
 Female 1.2 0.42–3.65 1.7 0.45–6.59 1.1 0.39–3.40 0.4 0.10–1.63
Ability to write
 Cannot write 1.0 1.0 1.0 1.0
 Can write 5.6** 1.25–25.29 8.1** 1.60–41.58 0.2*** 0.07–0.55 0.1*** 0.04–0.48
Physically active
 Not at all/not very 1.0 1.0 1.0 1.0
 Fairly/very 1.7 0.60–4.59 1.4 0.47–4.24 0.3** 0.12–0.91 0.3* 0.13–1.07
Relationship of carer to respondent
 Spouse 1.0 1.0 1.0 1.0
 Child/grandchild 1.3 0.38–4.81 1.3 0.27–6.01 2.8 0.57–14.20 2.8 0.48–16.09
 Other relative 0.5 0.05–4.97 0.5 0.05–5.20 2.3 0.30–17.89 2.9 0.31–26.37
 Neighbour 1.4 0.22–8.25 1.0 0.14–7.58 2.8 0.36–22.32 5.3 0.55–51.42
 Other 3.4 0.78–14.50 3.3 0.68–16.45 7.1** 1.23–41.25 12.2** 1.73–85.89
N 150 150
Pseudo R2 0.13 0.20

Notes: OR = odds ratio; CI = confidence interval.

a Crude OR shows results for the variable without other controls. b Adjusted OR shows results for the full model. * p < 0.1; ** p < 0.05; *** p < 0.01. Results which are statistically significant at the 10 per cent level are shown in bold.

Steffener et al (2016) have shown that education and physical activity are linked to lower brain age, and Kwak et al (2018) have demonstrated that brain age and subjective age are linked. We have already shown that the older people in our sample do not necessarily possess a firm knowledge of their precise date of birth and, therefore, their exact chronological age, and it is possible that their reporting of their age was influenced by their health and mental capacity. The results here demonstrate that even if this is the case, the physical capacity and literacy of an older person influence the age allocated them by their carer even more than they influence the age given by the older person themself.

Discussion

This article set out to examine the accuracy of age reporting by and for older people in Uganda, and to assess influences on the ages reported. We have demonstrated that age reporting among older people in Uganda is rough or very rough, and therefore that little store should be put on the precise age individuals report.

Our analysis contributes to the debate about how to interpret the presence of age heaping. Our finding that literate people were more likely to report their age as a round number casts doubt on the common assumption that rough age reporting is due to the inability to keep track of records and certificates or to calculate age from those documents (Mba, 2014; Lyons-Amos and Stones, 2017; Tollnek and Baten, 2017; Francis et al, 2019). Although more literate people were more likely to report an age heaped on 0 or 5, that age was more likely to be consistent with the date of birth that they returned than was the case for illiterate people, confirming that rough reporting does not indicate innumeracy or low human capital. We suggest that our finding is linked to the introduction of National ID cards in Uganda in 2014 (five years before our survey took place) and the possibility that for many people, as such cards represented the first time they possessed a document with their date of birth, the date was derived from a rounded age in 2014. Although we disagree that age heaping is an accurate indicator of numeracy, correlation between the two are common because the development of education and numeracy has often gone hand in hand with the introduction of registration and the growing requirement to report age or date of birth. However for today’s older generations in Uganda, and probably those in other societies too, we should be wary of interpreting age heaping as an indicator of low human capital.

Our theory needs further testing with a larger sample and an investigation of the link between the date of receipt of ID cards and literacy. However it strongly indicates that age reporting is a better measure of administrative modernisation, as argued by A’Hearn et al (2022a) and Spenneman (2017), than of human capital, and that the registration of births and issuing of birth certificates is likely to be crucial. The introduction of ID registration later in people’s lives can result in the assigning of inaccurate birth dates, which may be rounded according to the year they are assigned. If large segments of the population go through ID registration at the same time, this can introduce patterns of age heaping among those segments. The current push for much-needed registration and ID documentation (Setel et al, 2007; Hunter, 2019) may not, therefore, be accompanied by immediate improvements in age reporting among older people.

Age heaping is easy to detect in survey data, but without reliable evidence of date of birth, it is very difficult to detect systematic under- or over-reporting of age. If people do not know their age, do not have a certificate to look it up or cannot read that certificate, the age they report might depend on their health, daily capabilities or social roles. In other words, it might be a better reflection of their subjective, perceived or felt age, or how they see their position in the world (Gilleard, 2022). Healthier people might underestimate their age and unhealthy people overestimate their age, and this bias would make it difficult to investigate the true links between chronological age and frailty. Although we are not able to prove that people’s reporting of their own age was influenced by their health and functional capabilities, we did find that these aspects affected the age that carers reported for older people. Carers who knew the older person less well were more likely to overestimate that person’s age, and discrepancies in age reporting by respondents and carers appear to be largely due to the perceptions of the carers: in other words the health, appearance or capability (including education level) of the older people affected how old their carers thought they were. We can draw three implications from this.

First, it means that proxy reports of age informed by health or appearance are highly likely to be systematically biased. Of course it might be the case, as found by Smit et al (1997), that certain forms of poor health (particularly where this is related to cognitive function) also reduce the consistency of age recall in respondents themselves. This sort of effect, although it will increase variability of results, is unlikely to produce a strong and important bias. In contrast, our finding of a correlation between physical health and the assessment of age by others has important repercussions regarding surveys where information for older people is frequently given by proxy informants (for example, by head of household). These findings are particularly important for comparisons of healthy life expectancies and differences between groups within places, between different places and over time, as different cultures and norms as well as underlying health status may affect assessments of age.

Second, there are implications for the use of other age detection methods, such as computer-aided age identification. A suite of emerging methods for validating and improving ages using computer vision (these methods use a dataset of validated ages to train a computer programme to predict age) have had success in distinguishing women under 50 from those over 50 (Helleringer et al, 2019). However poor health, ever smoking and prolonged sun exposure were associated with large errors, and these may be the same sort of errors which affect human estimation of age (Helleringer et al, 2019). This suggests that computer vision may still be some way off producing accurate reflections of age among older people.

Third, our research raises questions about the definition of old age. Our findings indicate that peri-urban societies in Uganda define old age according to social role or capability more than birth dates – this is unsurprising given the lack of birth registration for this cohort. It is possible that this will change in younger generations as more schooling and more heavily documented lives repeatedly reinforce individuals’ knowledge of their own age and provide them with the documentation to check and prove it. At the same time, our findings remind us that chronological age – although crucial from the perspective of population modelling and forecasting – is not always a good indicator of biological age, which may be more important (although difficult to measure) for defining vulnerability and related interventions (Jylhävä et al, 2017).

Our study has a few limitations. Primary among these is the very small sample size, which is due to the pilot nature of the study and limited funding, but the indicative responses still provide useful evidence, suggest that a larger study would be very fruitful, and can be used to inform the size of such a study. It is also important to consider the possibility of bias due to non-response and coverage. Non-response rates in the study were low, with only three individuals not interviewed – this was because of frailty. Possibly more worrying is that the study sample was based on self-reports of age – we only interviewed people who claimed to be 60 or older, and therefore those who underestimated their real age will have been excluded. Some analyses have argued that 50 is a more appropriate cut-off point for ‘old age’ in Africa, where life expectancy is still relatively low (Velkoff and Kowal, 2007; Ferreira and Kowal, 2013), but as we only had resources to recruit a small sample, we decided to limit our study to people reporting their age as 60 and over; if we had included those reporting ages of 50 to 59, the number of respondents in our study with higher ages would have been severely limited. A large study in the future would benefit from including a broader age range. Because our focus was on the accuracy of reporting of chronological age, we did not ask how old people felt they were, but it would be interesting to include this in a future study. Although our study highlighted some of the possible implications of a lack of ID documentation and inaccurate information on documents, it did not address these issues in detail. This would be a fruitful avenue for future research, using a qualitative approach to gather perspectives from older people themselves as well as analysing quantitative data.

Conclusion

It is difficult, if not impossible, to conduct research on ageing without relying on reported age – if only to identify older people. The accuracy of reported age is rarely questioned, but we have demonstrated that when many older people in Uganda report their age, they may be reporting something closer to their felt age than their chronological age. And when others report the age of older people, they are liable to use the person’s appearance or functional capability as a guide, leading to systematic biases in proxy reporting. Data on felt age may be just as valuable as data on chronological age, particularly when planning services, but it should not automatically be assumed that reported age accurately reflects chronological age. To report their chronological age correctly, a person needs: (1) to have an accurate record of their birth date; (2) to be able to read that record; and (3) to be able to calculate their age from their birth date and the current date. The presence of age misreporting has generally been interpreted as indicating deficiencies in literacy (2) or numeracy (3), but in our sample, age heaping had little relationship with the ability to calculate age from birth date and current date, and was associated with higher literacy rather than lower. We argue that the possession of a document that is issued at or near date of birth (such as birth or baptism registration) and which shows that date is likely to be a prerequisite for accurate knowledge of age. Literate and numerate people may rely on these, or documents gained later in life which record a date of birth, for knowledge of their age. However, the dates of birth on documents gained later may reflect an estimated age, and where many people gain such documents at the same time – such as during the widespread introduction of national ID registration in Uganda – heaping on particular years of birth may be introduced into the population. In relation to places where relatively few people have birth certificates, therefore, both those carrying out studies of older people and those developing health and social policies should be aware of potential age misreporting due to either respondents’ hazy knowledge of their own age or issues related to proxy age reporting, and the biases that this entails.

As estimated in 2021, globally, around 850 million people – around one in every nine at that time – do not have an official proof of ID, which often prevents them from accessing services and fulfilling rights (World Bank, 2021). Nearly all of these live in low- and lower-middle-income countries, and over half are in sub-Saharan Africa. Lack of documents such as birth certificates remains a barrier to ID registration (World Bank, 2021), but despite a push for improving documentation and birth registration, a recent survey estimates that globally around 30 per cent of infants have not been registered and 40 per cent do not possess a birth certificate (UNICEF, 2019). Together with our findings, these facts imply that many children born today may not have a firm grasp of their chronological age when they reach older ages. Where birth registration is introduced, it is likely to be accessed first by wealthier sections of populations, and this will potentially generate differentials in age reporting in the future. In these circumstances, the ability to gain an ID document in later life should not depend on the production of a birth certificate, even though that may result in the assigning of an inaccurate birth date.

Notes

1

For example, in the Namibian census of 1991, 2001 and 2011, there was heaping on ages ending in 1 (Shipanga and Shinyemba, 2023), and in the Malawi census of 1998 and 2008, there was heaping on ages ending in 8 (Fajardo-González et al, 2014).

2

Unlike in wealthier regions, Africa has few surveys directed at older people, and most demographic data comes from censuses, nationally representative surveys (such as the DHS) and a small number of location-specific DSSs.

3

Following the questionnaire, some participants were invited to participate in focus groups to discuss what it means to be old, a team of artists created art works based on the older people, and a combination of artists and local nongovernmental organisations led co-production art workshops with the older people. These aspects were reported on by other teams within the project, as were detailed analyses of the health of the older people.

4

The normal Whipple Index starts with age 23 and ends with age 62 in order to capture both people who round their age up (eg from 23 or 24 to 25) and those who round their age down (from 61 or 62 to 60). The majority of people who round their ages are therefore captured in both the numerator and the denominator of the calculation. Starting with age 60 means that those aged 58 and 59 who reported themselves as 60 will be included in numerator of the calculation but not the denominator and this leads to over-estimation of the index. Because the population declines rapidly in old age this is not balanced by people in the oldest ages (eg ages 96 and 97 or ages 101 and 102) rounding down.

5

All but three participants reported a plausible birth date.

6

Of course there will be some people whose birthdays really did fall on the first of the month, and indeed on the first of January. We cannot tell the difference between these people and those who did not specify a day or month and thus the percentages of people reporting with medium and low precision will be slightly over-estimated.

7

Although we searched in parish registers for the baptism records of the respondents, poor survival of records, lack of access and migration meant that we were only able to identify nine baptism records this way.

8

Ideally, such a study would be undertaken in a region with a majority of Christians, relatively low migration and a single, long-established parish with a tradition of birth (as well as baptism) date recording (Helleringer et al, 2019).

Funding

This research was funded the UKRI Arts and Humanities Research Council (AHRC) as part of a joint Arts and Humanities (AHRC) and Medical Research Council (MRC) Global Challenges Research Fund (GCRF) (grant reference AH/R005990/1). For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising. The funders had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

Acknowledgements

We are grateful to the older persons and their carers who participated in the study. We acknowledge support from the district and local Council officials who gave administrative clearance to all the research assistants engaged in the data collection. We thank the churches that allowed us to search their baptism records. We are especially grateful to the other members of our project who supported, listened, advised and helped us grapple with the interdisciplinary nature of the project, particularly Carol Brayne, Noeline Nakasujja, Rosalind Parkes-Ratanshi, Louise Lafortune, Laureen Kahunde and Tennie Videler.

Ethical approval

Ethical approval was granted by Makerere School of Medicine Research Ethics Committee (SOMREC 2018-098), by Uganda National Council for Science and Technology (UNCST: SS246ES) and by the University of Cambridge.

Data availability

Data and materials will be archived at the University of Cambridge repository.

Conflict of interest

The authors declare that there is no conflict of interest.

References

  • A’Hearn, B., Baten, J. and Crayen, D. (2009) Quantifying quantitative literacy: age heaping and the history of human capital, The Journal of Economic History, 69(3): 783808. doi: 10.1017/S0022050709001120

    • Search Google Scholar
    • Export Citation
  • A’Hearn, B., Delfino, A. and Nuvolari, A. (2022a) Age heaping and its discontents: a response to Baten, Benati, and Ferber, The Economic History Review, 75(3): 97280. doi: 10.1111/ehr.13140

    • Search Google Scholar
    • Export Citation
  • A’Hearn, B., Delfino, A. and Nuvolari, A. (2022b) Rethinking age heaping: a cautionary tale from nineteenth‐century Italy, The Economic History Review, 75(1): 11137. doi: 10.1111/ehr.13087

    • Search Google Scholar
    • Export Citation
  • Aboderin, I. (2010) Understanding and advancing the health of older populations in sub-Saharan Africa: policy perspectives and evidence needs, Public Health Reviews, 32(2): 35776. doi: 10.1007/bf03391607

    • Search Google Scholar
    • Export Citation
  • Baten, J., Benati, G. and Ferber, S. (2022) Rethinking age heaping again for understanding its possibilities and limitations, The Economic History Review, 75(3): 96071. doi: 10.1111/ehr.13139

    • Search Google Scholar
    • Export Citation
  • Bordone, V., Arpino, B. and Rosina, A. (2020) Forever young? An analysis of the factors influencing perceptions of ageing, Ageing and Society, 40(8): 166993. doi: 10.1017/s0144686x19000084

    • Search Google Scholar
    • Export Citation
  • Cappelli, G. and Baten, J. (2021) Numeracy development in Africa: new evidence from a long-term perspective (1730–1970), Journal of Development Economics, 150: 102630. doi: 10.1016/j.jdeveco.2021.102630

    • Search Google Scholar
    • Export Citation
  • Census of England and Wales (1881) Volume IV. General Report (1883) BPP 1883 LXXX.583. London.

  • Christensen, K., Thinggaard, M., McGue, M., Rexbye, H., Hjelmborg, J.v.B., Aviv, A., Gunn, D., van der Ouderaa, F. and Vaupel, J.W. (2009) Perceived age as clinically useful biomarker of ageing: cohort study, BMJ, 339: b5262. doi: 10.1136/bmj.b5262

    • Search Google Scholar
    • Export Citation
  • Coale, A.J. and Kisker, E.E. (1986) Mortality crossovers: reality or bad data? Population Studies, 40(3): 389401. doi: 10.1080/0032472031000142316

    • Search Google Scholar
    • Export Citation
  • Coale, A.J. and Li, S.M. (1991) The effect of age misreporting in China on the calculation of mortality rates at very high ages, Demography, 28(2): 293301. doi: 10.2307/2061281

    • Search Google Scholar
    • Export Citation
  • Crayen, D. and Baten, J. (2010) Global trends in numeracy 1820–1949 and its implications for long-term growth, Explorations in Economic History, 47(1): 8299. doi: 10.1016/j.eeh.2009.05.004

    • Search Google Scholar
    • Export Citation
  • Demir Erbil, D. and Hazer, O. (2022) Determination of the factors affecting age perception: the case of Turkey, Educational Gerontology, 48(9): 42944. doi: 10.1080/03601277.2022.2045737

    • Search Google Scholar
    • Export Citation
  • Elo, I.T. and Preston, S.H. (1994) Estimating African-American mortality from inaccurate data, Demography, 31(3): 42758. doi: 10.2307/2061751

    • Search Google Scholar
    • Export Citation
  • Elo, I.T., Preston, S.H., Rosenwaike, I., Hill, M. and Cheney, T.P. (1996) Consistency of age reporting on death certificates and Social Security records among elderly African Americans, Social Science Research, 25(3): 292307. doi: 10.1006/ssre.1996.0014

    • Search Google Scholar
    • Export Citation
  • Ewbank, D.C. (1981) Age Misreporting and Age-Selective Underenumeration: Sources, Patterns, and Consequences for Demographic Analysis, Vol. 4, Washington, DC: National Academy Press.

    • Search Google Scholar
    • Export Citation
  • Fajardo-González, J., Attanasio, L. and Ha, J.T. (2014) An assessment of the age reporting in the IPUMS-I microdata, Paper submitted for presentation at the 2014 Annual Meeting of the Population Association of America, https://paa2014.populationassociation.org/papers/140099.

    • Search Google Scholar
    • Export Citation
  • Fayehun, O., Ajayi, A.I., Onuegbu, C. and Egerson, D. (2020) Age heaping among adults in Nigeria: evidence from the Nigeria Demographic and Health Surveys 2003–2013, Journal of Biosocial Science, 52(1): 1329. doi: 10.1017/S0021932019000348

    • Search Google Scholar
    • Export Citation
  • Ferreira, M. and Kowal, P. (2013) A minimum data set on ageing and older persons in sub-Saharan Africa: process and outcome, African Population Studies, 21(1): 1936. doi: 10.11564/21-1-348

    • Search Google Scholar
    • Export Citation
  • Földvári, P., Van Leeuwen, B. and Van Leeuwen-Li, J. (2012) How did women count? A note on gender-specific age heaping differences in the sixteenth to nineteenth centuries, The Economic History Review, 65(1): 30413. doi: 10.1111/j.1468-0289.2010.00582.x

    • Search Google Scholar
    • Export Citation
  • Francis, E.-B., Senyefia, B.-A. and Borbor, B.S. (2019) Statistical analysis of age reporting errors by insured and uninsured patients in Cape Coast Teaching Hospital of Ghana, Biomedical Statistics and Informatics, 4(2): 1521. doi: 10.11648/j.bsi.20190402.11

    • Search Google Scholar
    • Export Citation
  • Gilleard, C. (2022) Age, subjectivity and the concept of subjective age: a critique, Journal of Aging Studies, 60: 101001. doi: 10.1016/j.jaging.2022.101001

    • Search Google Scholar
    • Export Citation
  • Handforth, C. and Wilson, M. (2019) Digital identity country profile: Uganda, https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2019/02/Digital-Identity-Country-Report-Uganda.pdf.

    • Search Google Scholar
    • Export Citation
  • Hausknecht, S., Low, L.-F., O’Loughlin, K., McNab, K. and Clemson, L. (2020) Older adults’ self-perceptions of aging and being older: a scoping review, The Gerontologist, 60(7): e524e534. doi: 10.1093/geront/gnz153

    • Search Google Scholar
    • Export Citation
  • Helleringer, S., You, C., Fleury, L., Douillot, L., Diouf, I., Ndiaye, C.T., Delaunay, V. and Vidal, R. (2019) Improving age measurement in low- and middle-income countries through computer vision: a test in Senegal, Demographic Research, 40: 219260. doi: 10.4054/demres.2019.40.9

    • Search Google Scholar
    • Export Citation
  • HelpAge International (2011) Challenges and Opportunities for Age Verification in Low- and Middle-Income Countries, Pension Watch: Briefings on Social Security in Older Age Briefing No 6, London: HelpAge International, www.helpage.org/silo/files/pension-watch-briefing-no-6--challenges-and-opportunites-for-age-verification.pdf.

    • Search Google Scholar
    • Export Citation
  • Hunter, W. (2019) Identity documents, welfare enhancement, and group empowerment in the Global South, The Journal of Development Studies, 55(3): 36683. doi: 10.1080/00220388.2018.1451637

    • Search Google Scholar
    • Export Citation
  • Johnson, P., Spoorenberg, T., Hertog, S. and Gerland, P. (2022) Method Protocol for the Evaluation of Census Population Data by Age and Sex, UN DESA/POP/2022/TP/No.5, New York: United Nations Department of Economic and Social Affairs, Population Division, www.un.org/development/desa/pd/sites/www.un.org.development.desa.pd/files/undesa_pd_2022_tp-methodprotocol.pdf.

    • Search Google Scholar
    • Export Citation
  • Jones, J.A.B., Nash, U.W., Vieillefont, J., Christensen, K., Misevic, D. and Steiner, U. (2019) The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100, Scientific Data, 6: 246. doi: 10.1038/s41597-019-0245-9

    • Search Google Scholar
    • Export Citation
  • Juif, D., Baten, J. and Pérez-Artés, M.C. (2020) Numeracy of religious minorities in Spain and Portugal during the Inquisition era, Revista de Historia Económica/Journal of Iberian and Latin American Economic History, 38(1): 14784. doi: 10.1017/s021261091900034x

    • Search Google Scholar
    • Export Citation
  • Jylhävä, J., Pedersen, N.L. and Hägg, S. (2017) Biological age predictors, EBioMedicine, 21: 2936. doi: 10.1016/j.ebiom.2017.03.046

    • Search Google Scholar
    • Export Citation
  • Kowal, P. and Dowd, J.E. (2001) Definition of an Older Person. Proposed Working Definition of an Older Person in Africa for the MDS Project, Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • Kwak, S., Kim, H., Chey, J. and Youm, Y. (2018) Feeling how old I am: subjective age is associated with estimated brain age, Frontiers in Aging Neuroscience, 10: 168. doi: 10.3389/fnagi.2018.00168

    • Search Google Scholar
    • Export Citation
  • Lankoandé, Y.B., Masquelier, B., Zabre, P., Bangré, H., Duthé, G., Soura, A.B., Pison, G. and Ali, S. (2022) Estimating mortality from census data: a record-linkage study of the Nouna Health and Demographic Surveillance System in Burkina Faso, Demographic Research, 46: 65380. doi: 10.4054/demres.2022.46.22

    • Search Google Scholar
    • Export Citation
  • Lyons-Amos, M. and Stones, T. (2017) Trends in Demographic and Health Survey data quality: an analysis of age heaping over time in 34 countries in sub Saharan Africa between 1987 and 2015, BMC Research Notes, 10: 760. doi: 10.1186/s13104-017-3091-x

    • Search Google Scholar
    • Export Citation
  • Macia, E., Duboz, P., Montepare, J.M. and Gueye, L. (2012) Age identity, self-rated health, and life satisfaction among older adults in Dakar, Senegal, European Journal of Ageing, 9(3): 24353. doi: 10.1007/s10433-012-0227-7

    • Search Google Scholar
    • Export Citation
  • Macia, E., Dial, F.B., Montepare, J.M., Hane, F. and Duboz, P. (2019) Ageing and the body: one African perspective, Ageing and Society, 39(4): 81535. doi: 10.1017/s0144686x17001313

    • Search Google Scholar
    • Export Citation
  • Malik, M.A. (2021) Age heaping pattern and data quality: evidence from Indian Household Survey Data (1991–2016), Communications in Statistics: Case Studies, Data Analysis and Applications, 7(3): 38293. doi: 10.1080/23737484.2021.1952492

    • Search Google Scholar
    • Export Citation
  • Mba, C.J. (2014) Examining the accuracy of age-sex data: an evaluation of recent sub-Saharan African population censuses, in C.O. Odimegwu and J. Kekovole (eds) Continuity and Change in Sub-Saharan African Demography, New York: Routledge, pp 1236.

    • Search Google Scholar
    • Export Citation
  • Morelock, J.C., Stokes, J.E. and Moorman, S.M. (2017) Rewriting age to overcome misaligned age and gender norms in later life, Journal of Aging Studies, 40: 1622. doi: 10.1016/j.jaging.2016.12.003

    • Search Google Scholar
    • Export Citation
  • Pardeshi, G. (2010) Age heaping and accuracy of age data collected during a community survey in the Yavatmal district, Maharashtra, Indian Journal of Community Medicine, 35(3): 3915. doi: 10.4103/0970-0218.69256

    • Search Google Scholar
    • Export Citation
  • Perrin, F. (2020) Gender gap in numeracy and the role of marital status: evidence from 19th century France, Revue d’économie politique, 130(1): 5176. doi: 10.3917/redp.301.0051

    • Search Google Scholar
    • Export Citation
  • Pinquart, M. and Wahl, H.-W. (2021) Subjective age from childhood to advanced old age: a meta-analysis, Psychology and Aging, 36(3): 394406. doi: 10.1037/pag0000600

    • Search Google Scholar
    • Export Citation
  • Preston, S.H., Elo, I.T., Rosenwaike, I. and Hill, M. (1996) African-American mortality at older ages: results of a matching study, Demography, 33(2): 193209. doi: 10.2307/2061872

    • Search Google Scholar
    • Export Citation
  • Randall, S. and Coast, E. (2016) The quality of demographic data on older Africans, Demographic Research, 34: 14374. doi: 10.4054/demres.2016.34.5

    • Search Google Scholar
    • Export Citation
  • Resilient Africa Network (2019) Understanding the Benefits, Costs, and Challenges of the National Identification System in Uganda. Findings from a Household Survey and a Costing Study, Kampala: Resilient Africa Network and Makerere University, https://tinyurl.com/2p9d64ps.

    • Search Google Scholar
    • Export Citation
  • Rosenzweig, S.C. (2021) Age is measured with systematic measurement error in developing country surveys: a diagnosis and analysis of consequences, Research & Politics, 8(3). doi: 10.1177/20531680211044068

    • Search Google Scholar
    • Export Citation
  • Sagner, A., Kowal, P. and Dowd, J.E. (2002) Defining ‘Old Age’: Markers of Old Age in Sub-Saharan Africa and the Implications for Cross-Cultural Research, The Minimum Data Set (MDS) Project on Ageing and Older Adults in sub-Saharan Africa Discussion Paper, No 1., WHO MDS Project. doi: 10.13140/2.1.2055.4885

    • Search Google Scholar
    • Export Citation
  • Schönstein, A., Ngo, D.T.T., Stephan, Y., Siè, A., Harling, G., Bärnighausen, T. and Wahl, H.‐W. (2021) Feeling younger in rural Burkina Faso: exploring the role of subjective age in the light of previous research from high-income countries, The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 76(10): 202940. doi: 10.1093/geronb/gbab151

    • Search Google Scholar
    • Export Citation
  • Setel, P.W., Macfarlane, S.B., Szreter, S., Mikkelsen, L., Jha, P., Stout, S. and AbouZahr, C. (2007) A scandal of invisibility: making everyone count by counting everyone, The Lancet, 370(9598): 156977. doi: 10.1016/S0140-6736(07)61307-5

    • Search Google Scholar
    • Export Citation
  • Shipanga, I. and Shinyemba, T. (2023) Evaluation of age-sex data collected in the 1991, 2001 and 2011 population and housing cnsuses of Namibia, Asian Journal of Population Sciences, 2(1): 116. doi: 10.3126/ajps.v2i1.51087

    • Search Google Scholar
    • Export Citation
  • Shryock, H.S. and Siegel, J.S. (1976) The Methods and Materials of Demography (condensed edn), New York: Academic Press.

  • Smit, J.H., Deeg, D.J.H. and Schmand, B.A. (1997) Asking the age question in elderly populations: a reverse record check study, The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 52B(4): P175P177. doi: 10.1093/geronb/52b.4.p175

    • Search Google Scholar
    • Export Citation
  • Spennemann, D.H.R. (2017) Age heaping among Indian Hawkers in South-Eastern Australia and their source communities in the Punjab, Journal of Sikh and Punjab Studies, 24(1–2): 149202.

    • Search Google Scholar
    • Export Citation
  • Spoorenberg, T. (2007) Quality of age reporting: extension and application of the modified Whipple’s index, Population (English Edition), 62(4): 72941. doi: 10.3917/pope.704.0729

    • Search Google Scholar
    • Export Citation
  • Steffener, J., Habeck, C., O’Shea, D., Razlighi, Q., Bherer, L. and Stern, Y. (2016) Differences between chronological and brain age are related to education and self-reported physical activity, Neurobiology of Aging, 40: 13844. doi: 10.1016/j.neurobiolaging.2016.01.014

    • Search Google Scholar
    • Export Citation
  • Szołtysek, M., Poniat, R. and Gruber, S. (2018) Age heaping patterns in Mosaic data, Historical Methods: A Journal of Quantitative and Interdisciplinary History, 51(1): 1338. doi: 10.1080/01615440.2017.1393359

    • Search Google Scholar
    • Export Citation
  • Tollnek, F. and Baten, J. (2014) Age-heaping-based human capital estimates, in C. Diebolt and M. Haupert (eds) Handbook of Cliometrics, Berlin: Springer, pp 13154.

    • Search Google Scholar
    • Export Citation
  • Tollnek, F. and Baten, J. (2017) Farmers at the heart of the ‘human capital revolution’? Decomposing the numeracy increase in early modern Europe, The Economic History Review, 70(3): 779809. doi: 10.1111/ehr.12382

    • Search Google Scholar
    • Export Citation
  • UNICEF (2019) Birth Registration for Every Child by 2030: Are We On Track? New York: UNICEF.

  • van der Straaten, J. (2022) Uganda’s sine qua non National ID – without which, not, SSRN. doi: 10.2139/ssrn.4106236

  • Velkoff, V.A. and Kowal, P.R. (2007) Population Aging in Sub-Saharan Africa: Demographic Dimensions 2006, Current Population Reports, P95/07-1, Washington, DC: US Government Printing Office, www.census.gov/library/publications/2007/demo/p95-07-1.html.

    • Search Google Scholar
    • Export Citation
  • Voelkle, M.C., Ebner, N.C., Lindenberger, U. and Riediger, M. (2012) Let me guess how old you are: effects of age, gender, and facial expression on perceptions of age, Psychology and Aging, 27(2): 26577. doi: 10.1037/a0025065

    • Search Google Scholar
    • Export Citation
  • Wak, G., Bangha, M., Azongo, D. Oduro, A. and Kwankye, S. (2017) Data reliability: comparison between Census and Health and Demographic Surveillance System (HDSS) outputs for Kassena-Nankana East and West Districts, Ghana, Population Review, 56(1): 3145. doi: 10.1353/prv.2017.0001

    • Search Google Scholar
    • Export Citation
  • West, K.K., Robinson, J.G. and Bentley, M. (2005) Did proxy respondents cause age heaping in the Census 2000? Paper presented at Joint Statistical Meetings: ASA Section on Survey Research Methods, Minneapolis.

  • World Bank (2021) ID4D Global Dataset – Volume 1: 2021: Global ID Coverage Estimates, Washington, DC: International Bank for Reconstruction and Development, World Bank Group, http://documents.worldbank.org/curated/en/099705012232226786/P176341032c1ef0b20adf10abad304425ef.

    • Search Google Scholar
    • Export Citation
  • Figure 1:

    Distribution of ages reported by respondents and carers

  • Figure 2:

    Precision of date of birth reported by respondents, by ability to write

  • Figure 3:

    Discrepancies in age reported by respondents and carers, by various indicators

  • A’Hearn, B., Baten, J. and Crayen, D. (2009) Quantifying quantitative literacy: age heaping and the history of human capital, The Journal of Economic History, 69(3): 783808. doi: 10.1017/S0022050709001120

    • Search Google Scholar
    • Export Citation
  • A’Hearn, B., Delfino, A. and Nuvolari, A. (2022a) Age heaping and its discontents: a response to Baten, Benati, and Ferber, The Economic History Review, 75(3): 97280. doi: 10.1111/ehr.13140

    • Search Google Scholar
    • Export Citation
  • A’Hearn, B., Delfino, A. and Nuvolari, A. (2022b) Rethinking age heaping: a cautionary tale from nineteenth‐century Italy, The Economic History Review, 75(1): 11137. doi: 10.1111/ehr.13087

    • Search Google Scholar
    • Export Citation
  • Aboderin, I. (2010) Understanding and advancing the health of older populations in sub-Saharan Africa: policy perspectives and evidence needs, Public Health Reviews, 32(2): 35776. doi: 10.1007/bf03391607

    • Search Google Scholar
    • Export Citation
  • Baten, J., Benati, G. and Ferber, S. (2022) Rethinking age heaping again for understanding its possibilities and limitations, The Economic History Review, 75(3): 96071. doi: 10.1111/ehr.13139

    • Search Google Scholar
    • Export Citation
  • Bordone, V., Arpino, B. and Rosina, A. (2020) Forever young? An analysis of the factors influencing perceptions of ageing, Ageing and Society, 40(8): 166993. doi: 10.1017/s0144686x19000084

    • Search Google Scholar
    • Export Citation
  • Cappelli, G. and Baten, J. (2021) Numeracy development in Africa: new evidence from a long-term perspective (1730–1970), Journal of Development Economics, 150: 102630. doi: 10.1016/j.jdeveco.2021.102630

    • Search Google Scholar
    • Export Citation
  • Census of England and Wales (1881) Volume IV. General Report (1883) BPP 1883 LXXX.583. London.

  • Christensen, K., Thinggaard, M., McGue, M., Rexbye, H., Hjelmborg, J.v.B., Aviv, A., Gunn, D., van der Ouderaa, F. and Vaupel, J.W. (2009) Perceived age as clinically useful biomarker of ageing: cohort study, BMJ, 339: b5262. doi: 10.1136/bmj.b5262

    • Search Google Scholar
    • Export Citation
  • Coale, A.J. and Kisker, E.E. (1986) Mortality crossovers: reality or bad data? Population Studies, 40(3): 389401. doi: 10.1080/0032472031000142316

    • Search Google Scholar
    • Export Citation
  • Coale, A.J. and Li, S.M. (1991) The effect of age misreporting in China on the calculation of mortality rates at very high ages, Demography, 28(2): 293301. doi: 10.2307/2061281

    • Search Google Scholar
    • Export Citation
  • Crayen, D. and Baten, J. (2010) Global trends in numeracy 1820–1949 and its implications for long-term growth, Explorations in Economic History, 47(1): 8299. doi: 10.1016/j.eeh.2009.05.004

    • Search Google Scholar
    • Export Citation
  • Demir Erbil, D. and Hazer, O. (2022) Determination of the factors affecting age perception: the case of Turkey, Educational Gerontology, 48(9): 42944. doi: 10.1080/03601277.2022.2045737

    • Search Google Scholar
    • Export Citation
  • Elo, I.T. and Preston, S.H. (1994) Estimating African-American mortality from inaccurate data, Demography, 31(3): 42758. doi: 10.2307/2061751

    • Search Google Scholar
    • Export Citation
  • Elo, I.T., Preston, S.H., Rosenwaike, I., Hill, M. and Cheney, T.P. (1996) Consistency of age reporting on death certificates and Social Security records among elderly African Americans, Social Science Research, 25(3): 292307. doi: 10.1006/ssre.1996.0014

    • Search Google Scholar
    • Export Citation
  • Ewbank, D.C. (1981) Age Misreporting and Age-Selective Underenumeration: Sources, Patterns, and Consequences for Demographic Analysis, Vol. 4, Washington, DC: National Academy Press.

    • Search Google Scholar
    • Export Citation
  • Fajardo-González, J., Attanasio, L. and Ha, J.T. (2014) An assessment of the age reporting in the IPUMS-I microdata, Paper submitted for presentation at the 2014 Annual Meeting of the Population Association of America, https://paa2014.populationassociation.org/papers/140099.

    • Search Google Scholar
    • Export Citation
  • Fayehun, O., Ajayi, A.I., Onuegbu, C. and Egerson, D. (2020) Age heaping among adults in Nigeria: evidence from the Nigeria Demographic and Health Surveys 2003–2013, Journal of Biosocial Science, 52(1): 1329. doi: 10.1017/S0021932019000348

    • Search Google Scholar
    • Export Citation
  • Ferreira, M. and Kowal, P. (2013) A minimum data set on ageing and older persons in sub-Saharan Africa: process and outcome, African Population Studies, 21(1): 1936. doi: 10.11564/21-1-348

    • Search Google Scholar
    • Export Citation
  • Földvári, P., Van Leeuwen, B. and Van Leeuwen-Li, J. (2012) How did women count? A note on gender-specific age heaping differences in the sixteenth to nineteenth centuries, The Economic History Review, 65(1): 30413. doi: 10.1111/j.1468-0289.2010.00582.x

    • Search Google Scholar
    • Export Citation
  • Francis, E.-B., Senyefia, B.-A. and Borbor, B.S. (2019) Statistical analysis of age reporting errors by insured and uninsured patients in Cape Coast Teaching Hospital of Ghana, Biomedical Statistics and Informatics, 4(2): 1521. doi: 10.11648/j.bsi.20190402.11

    • Search Google Scholar
    • Export Citation
  • Gilleard, C. (2022) Age, subjectivity and the concept of subjective age: a critique, Journal of Aging Studies, 60: 101001. doi: 10.1016/j.jaging.2022.101001

    • Search Google Scholar
    • Export Citation
  • Handforth, C. and Wilson, M. (2019) Digital identity country profile: Uganda, https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2019/02/Digital-Identity-Country-Report-Uganda.pdf.

    • Search Google Scholar
    • Export Citation
  • Hausknecht, S., Low, L.-F., O’Loughlin, K., McNab, K. and Clemson, L. (2020) Older adults’ self-perceptions of aging and being older: a scoping review, The Gerontologist, 60(7): e524e534. doi: 10.1093/geront/gnz153

    • Search Google Scholar
    • Export Citation
  • Helleringer, S., You, C., Fleury, L., Douillot, L., Diouf, I., Ndiaye, C.T., Delaunay, V. and Vidal, R. (2019) Improving age measurement in low- and middle-income countries through computer vision: a test in Senegal, Demographic Research, 40: 219260. doi: 10.4054/demres.2019.40.9

    • Search Google Scholar
    • Export Citation
  • HelpAge International (2011) Challenges and Opportunities for Age Verification in Low- and Middle-Income Countries, Pension Watch: Briefings on Social Security in Older Age Briefing No 6, London: HelpAge International, www.helpage.org/silo/files/pension-watch-briefing-no-6--challenges-and-opportunites-for-age-verification.pdf.

    • Search Google Scholar
    • Export Citation
  • Hunter, W. (2019) Identity documents, welfare enhancement, and group empowerment in the Global South, The Journal of Development Studies, 55(3): 36683. doi: 10.1080/00220388.2018.1451637

    • Search Google Scholar
    • Export Citation
  • Johnson, P., Spoorenberg, T., Hertog, S. and Gerland, P. (2022) Method Protocol for the Evaluation of Census Population Data by Age and Sex, UN DESA/POP/2022/TP/No.5, New York: United Nations Department of Economic and Social Affairs, Population Division, www.un.org/development/desa/pd/sites/www.un.org.development.desa.pd/files/undesa_pd_2022_tp-methodprotocol.pdf.

    • Search Google Scholar
    • Export Citation
  • Jones, J.A.B., Nash, U.W., Vieillefont, J., Christensen, K., Misevic, D. and Steiner, U. (2019) The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100, Scientific Data, 6: 246. doi: 10.1038/s41597-019-0245-9

    • Search Google Scholar
    • Export Citation
  • Juif, D., Baten, J. and Pérez-Artés, M.C. (2020) Numeracy of religious minorities in Spain and Portugal during the Inquisition era, Revista de Historia Económica/Journal of Iberian and Latin American Economic History, 38(1): 14784. doi: 10.1017/s021261091900034x

    • Search Google Scholar
    • Export Citation
  • Jylhävä, J., Pedersen, N.L. and Hägg, S. (2017) Biological age predictors, EBioMedicine, 21: 2936. doi: 10.1016/j.ebiom.2017.03.046

    • Search Google Scholar
    • Export Citation
  • Kowal, P. and Dowd, J.E. (2001) Definition of an Older Person. Proposed Working Definition of an Older Person in Africa for the MDS Project, Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • Kwak, S., Kim, H., Chey, J. and Youm, Y. (2018) Feeling how old I am: subjective age is associated with estimated brain age, Frontiers in Aging Neuroscience, 10: 168. doi: 10.3389/fnagi.2018.00168

    • Search Google Scholar
    • Export Citation
  • Lankoandé, Y.B., Masquelier, B., Zabre, P., Bangré, H., Duthé, G., Soura, A.B., Pison, G. and Ali, S. (2022) Estimating mortality from census data: a record-linkage study of the Nouna Health and Demographic Surveillance System in Burkina Faso, Demographic Research, 46: 65380. doi: 10.4054/demres.2022.46.22

    • Search Google Scholar
    • Export Citation
  • Lyons-Amos, M. and Stones, T. (2017) Trends in Demographic and Health Survey data quality: an analysis of age heaping over time in 34 countries in sub Saharan Africa between 1987 and 2015, BMC Research Notes, 10: 760. doi: 10.1186/s13104-017-3091-x

    • Search Google Scholar
    • Export Citation
  • Macia, E., Duboz, P., Montepare, J.M. and Gueye, L. (2012) Age identity, self-rated health, and life satisfaction among older adults in Dakar, Senegal, European Journal of Ageing, 9(3): 24353. doi: 10.1007/s10433-012-0227-7

    • Search Google Scholar
    • Export Citation
  • Macia, E., Dial, F.B., Montepare, J.M., Hane, F. and Duboz, P. (2019) Ageing and the body: one African perspective, Ageing and Society, 39(4): 81535. doi: 10.1017/s0144686x17001313

    • Search Google Scholar
    • Export Citation
  • Malik, M.A. (2021) Age heaping pattern and data quality: evidence from Indian Household Survey Data (1991–2016), Communications in Statistics: Case Studies, Data Analysis and Applications, 7(3): 38293. doi: 10.1080/23737484.2021.1952492

    • Search Google Scholar
    • Export Citation
  • Mba, C.J. (2014) Examining the accuracy of age-sex data: an evaluation of recent sub-Saharan African population censuses, in C.O. Odimegwu and J. Kekovole (eds) Continuity and Change in Sub-Saharan African Demography, New York: Routledge, pp 1236.

    • Search Google Scholar
    • Export Citation
  • Morelock, J.C., Stokes, J.E. and Moorman, S.M. (2017) Rewriting age to overcome misaligned age and gender norms in later life, Journal of Aging Studies, 40: 1622. doi: 10.1016/j.jaging.2016.12.003

    • Search Google Scholar
    • Export Citation
  • Pardeshi, G. (2010) Age heaping and accuracy of age data collected during a community survey in the Yavatmal district, Maharashtra, Indian Journal of Community Medicine, 35(3): 3915. doi: 10.4103/0970-0218.69256

    • Search Google Scholar
    • Export Citation
  • Perrin, F. (2020) Gender gap in numeracy and the role of marital status: evidence from 19th century France, Revue d’économie politique, 130(1): 5176. doi: 10.3917/redp.301.0051

    • Search Google Scholar
    • Export Citation
  • Pinquart, M. and Wahl, H.-W. (2021) Subjective age from childhood to advanced old age: a meta-analysis, Psychology and Aging, 36(3): 394406. doi: 10.1037/pag0000600

    • Search Google Scholar
    • Export Citation
  • Preston, S.H., Elo, I.T., Rosenwaike, I. and Hill, M. (1996) African-American mortality at older ages: results of a matching study, Demography, 33(2): 193209. doi: 10.2307/2061872

    • Search Google Scholar
    • Export Citation
  • Randall, S. and Coast, E. (2016) The quality of demographic data on older Africans, Demographic Research, 34: 14374. doi: 10.4054/demres.2016.34.5

    • Search Google Scholar
    • Export Citation
  • Resilient Africa Network (2019) Understanding the Benefits, Costs, and Challenges of the National Identification System in Uganda. Findings from a Household Survey and a Costing Study, Kampala: Resilient Africa Network and Makerere University, https://tinyurl.com/2p9d64ps.

    • Search Google Scholar
    • Export Citation
  • Rosenzweig, S.C. (2021) Age is measured with systematic measurement error in developing country surveys: a diagnosis and analysis of consequences, Research & Politics, 8(3). doi: 10.1177/20531680211044068

    • Search Google Scholar
    • Export Citation
  • Sagner, A., Kowal, P. and Dowd, J.E. (2002) Defining ‘Old Age’: Markers of Old Age in Sub-Saharan Africa and the Implications for Cross-Cultural Research, The Minimum Data Set (MDS) Project on Ageing and Older Adults in sub-Saharan Africa Discussion Paper, No 1., WHO MDS Project. doi: 10.13140/2.1.2055.4885

    • Search Google Scholar
    • Export Citation
  • Schönstein, A., Ngo, D.T.T., Stephan, Y., Siè, A., Harling, G., Bärnighausen, T. and Wahl, H.‐W. (2021) Feeling younger in rural Burkina Faso: exploring the role of subjective age in the light of previous research from high-income countries, The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 76(10): 202940. doi: 10.1093/geronb/gbab151

    • Search Google Scholar
    • Export Citation
  • Setel, P.W., Macfarlane, S.B., Szreter, S., Mikkelsen, L., Jha, P., Stout, S. and AbouZahr, C. (2007) A scandal of invisibility: making everyone count by counting everyone, The Lancet, 370(9598): 156977. doi: 10.1016/S0140-6736(07)61307-5

    • Search Google Scholar
    • Export Citation
  • Shipanga, I. and Shinyemba, T. (2023) Evaluation of age-sex data collected in the 1991, 2001 and 2011 population and housing cnsuses of Namibia, Asian Journal of Population Sciences, 2(1): 116. doi: 10.3126/ajps.v2i1.51087

    • Search Google Scholar
    • Export Citation
  • Shryock, H.S. and Siegel, J.S. (1976) The Methods and Materials of Demography (condensed edn), New York: Academic Press.

  • Smit, J.H., Deeg, D.J.H. and Schmand, B.A. (1997) Asking the age question in elderly populations: a reverse record check study, The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 52B(4): P175P177. doi: 10.1093/geronb/52b.4.p175

    • Search Google Scholar
    • Export Citation
  • Spennemann, D.H.R. (2017) Age heaping among Indian Hawkers in South-Eastern Australia and their source communities in the Punjab, Journal of Sikh and Punjab Studies, 24(1–2): 149202.

    • Search Google Scholar
    • Export Citation
  • Spoorenberg, T. (2007) Quality of age reporting: extension and application of the modified Whipple’s index, Population (English Edition), 62(4): 72941. doi: 10.3917/pope.704.0729

    • Search Google Scholar
    • Export Citation
  • Steffener, J., Habeck, C., O’Shea, D., Razlighi, Q., Bherer, L. and Stern, Y. (2016) Differences between chronological and brain age are related to education and self-reported physical activity, Neurobiology of Aging, 40: 13844. doi: 10.1016/j.neurobiolaging.2016.01.014

    • Search Google Scholar
    • Export Citation
  • Szołtysek, M., Poniat, R. and Gruber, S. (2018) Age heaping patterns in Mosaic data, Historical Methods: A Journal of Quantitative and Interdisciplinary History, 51(1): 1338. doi: 10.1080/01615440.2017.1393359

    • Search Google Scholar
    • Export Citation
  • Tollnek, F. and Baten, J. (2014) Age-heaping-based human capital estimates, in C. Diebolt and M. Haupert (eds) Handbook of Cliometrics, Berlin: Springer, pp 13154.

    • Search Google Scholar
    • Export Citation
  • Tollnek, F. and Baten, J. (2017) Farmers at the heart of the ‘human capital revolution’? Decomposing the numeracy increase in early modern Europe, The Economic History Review, 70(3): 779809. doi: 10.1111/ehr.12382

    • Search Google Scholar
    • Export Citation
  • UNICEF (2019) Birth Registration for Every Child by 2030: Are We On Track? New York: UNICEF.

  • van der Straaten, J. (2022) Uganda’s sine qua non National ID – without which, not, SSRN. doi: 10.2139/ssrn.4106236

  • Velkoff, V.A. and Kowal, P.R. (2007) Population Aging in Sub-Saharan Africa: Demographic Dimensions 2006, Current Population Reports, P95/07-1, Washington, DC: US Government Printing Office, www.census.gov/library/publications/2007/demo/p95-07-1.html.

    • Search Google Scholar
    • Export Citation
  • Voelkle, M.C., Ebner, N.C., Lindenberger, U. and Riediger, M. (2012) Let me guess how old you are: effects of age, gender, and facial expression on perceptions of age, Psychology and Aging, 27(2): 26577. doi: 10.1037/a0025065

    • Search Google Scholar
    • Export Citation
  • Wak, G., Bangha, M., Azongo, D. Oduro, A. and Kwankye, S. (2017) Data reliability: comparison between Census and Health and Demographic Surveillance System (HDSS) outputs for Kassena-Nankana East and West Districts, Ghana, Population Review, 56(1): 3145. doi: 10.1353/prv.2017.0001

    • Search Google Scholar
    • Export Citation
  • West, K.K., Robinson, J.G. and Bentley, M. (2005) Did proxy respondents cause age heaping in the Census 2000? Paper presented at Joint Statistical Meetings: ASA Section on Survey Research Methods, Minneapolis.

  • World Bank (2021) ID4D Global Dataset – Volume 1: 2021: Global ID Coverage Estimates, Washington, DC: International Bank for Reconstruction and Development, World Bank Group, http://documents.worldbank.org/curated/en/099705012232226786/P176341032c1ef0b20adf10abad304425ef.

    • Search Google Scholar
    • Export Citation
Alice Reid University of Cambridge, UK

Search for other papers by Alice Reid in
Current site
Google Scholar
Close
,
Allen Kabagenyi University of Makerere, Uganda

Search for other papers by Allen Kabagenyi in
Current site
Google Scholar
Close
,
Stephen Ojiambo Wandera University of Makerere, Uganda

Search for other papers by Stephen Ojiambo Wandera in
Current site
Google Scholar
Close
,
Joshua Beinomugisha University of Makerere, Uganda

Search for other papers by Joshua Beinomugisha in
Current site
Google Scholar
Close
, and
Sarah Walters London School of Hygiene and Tropical Medicine, UK

Search for other papers by Sarah Walters in
Current site
Google Scholar
Close

Content Metrics

May 2022 onwards Past Year Past 30 Days
Abstract Views 28 28 0
Full Text Views 248 248 120
PDF Downloads 199 199 97

Altmetrics

Dimensions