Assessing the reliability of the 1986 and 1996 
Lesotho census data 
CHUKS J. MBA1 
ABSTRACT 
The present study attempts to assess the quality of reported age-sex 
distributions of Lesotho's 1986 and 1996 censuses using conventional 
demographic techniques. First, the data presented in single 
years are examined to identify patterns of digit preference and then 
Whipple's, Myers' and Bachi's indexes are computed in order to 
furnish a clear picture of the magnitude of deficiencies that might 
have occurred. The findings show that there is a tendency to prefer 
even numbers and avoid odd numbers in census enumeration in 
Lesotho. Also, the Whipple's index declined from 115 in 1986 to 106 
in 1996, while the Myers' and Bachi's indexes respectively declined 
from 11 and 7 in 1986 to 9 and 6. These results suggest that there 
are deficiencies in the Lesotho data. A closer assessment of the data 
is undertaken by curtailing part of the erratic fluctuations in single 
year age distribution via grouping the data in quinary ages and 
applying the age—sex accuracy index. The index yields a value of 35 
for the 1986 census and 32 for the 1996 census. The findings show 
that, though there are some distortions in the reported age—sex 
distributions of Lesotho, the data are fairly accurate and point to a 
modest improvement in quality over the decade. 
Introduction 
Demographic variables, like many social phenomena, are not easy to 
measure and control. This is mainly because of their continuous change 
and the statistical fluctuations arising from the unpredictability of human 
1 United Nations Regional Institute for Population Studies, 
University of Ghana, Legon, Ghana 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 111
behaviour, together with the varied extraneous factors which may be 
physical, economic, social and cultural, that influence and often 
determine the levels, patterns and forms of demographic development 
and processes (Bachi 1951, Keyfitz and Hieger 1990,Swamyetal. 1992). 
Furthermore, ease of measurement is hampered by the complex 
interrelationships among demographic variables that are also intricately 
related to other non-demographic factors. This situation normally 
complicates the difficulty with which planners and demographers are 
able to convince others of the importance of programmes aimed at 
achieving certain demographic targets. In addition, in the African context 
specifically there is the problem of the scarcity and quality of 
demographic data. Even with the recent progress in establishing and 
promoting survey programmes, notably under the World Fertility Survey 
and Demographic and Health Survey networks, gaps still remain (El- 
Attar 1998, Cleland 1996, Krotki 1978). Some countries have not had 
regular censuses and surveys; for example, Nigeria conducted a 
successful and acceptable national census only in 1991, a gap of 28 
years after the officially-endorsed one of 1963 (National Population 
Commission 1998). Also the coverage of demographic variables during 
enumerations has not been entire and adequate and, more often than 
not, data collected in the African region have content as well as 
enumeration errors which mar their quality and reliability for planning 
purposes (Cleland 19%, Sembajwe 1990, CaMwell 1966). 
One of the key questions that confront the social scientist using a 
particular data set for research is: what is the nature and magnitude of 
errors inherent in the data collected, with specific reference to the 
variables of interest? This question is consistent with the viewpoint of 
Spiegelman, who contends that one of the fundamental precautions 
that must be taken before embarking on the analysis and interpretation 
of demographic data is that "the quality of the observed data should be 
ascertained ...." (Spiegelman 1968:6). The thrust of this study is an 
attempt to answer this basic question with respect to the Kingdom of 
Lesotho, since the usefulness and reliability of demographic parameters 
derived from censuses depend on the quality of the data collected. 
Since age is critical to most demographic investigations, attention will 
112 JOURNAL OF SOCIAL DEVELOPMENT. IN AFRICA VCH. 18 NO 1 JANUARY 20O3
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
be focused primarily on the quality of age data. In this study, attention 
is primarily focused on the 1986 and 19% censuses, chiefly because the 
data are available and recent and because they are comparable in terms 
of content and coverage. 
Background information 
Lesotho, formerly Basutoland, attained political independence from 
Britain in 1966. The country occupies an area of 30,355 square kilometres, 
of which less than one-fifth is arable (Kingdom of Lesotho 1966). It is 
landlocked and wholly enclosed by the Republic of South Africa. About 
two-thirds of Lesotho consists of rugged, mountainous terrain that is' 
difficult to reach (Gill 1993, Kingdom of Lesotho 1966). The country is 
composed of ten administrative districts, with Maseru as the capital 
and 60 constituencies. 
Although Lesotho is politically free, it is economically dependent on 
South Africa, perhaps due to historical, geographical and physical 
reasons. South Africa determines Lesotho's wages and prices, its interest 
rates and customs receipts and the exchange rate of its monetary unit, 
as well as controlling all trade and communication links between Lesotho 
and the rest of the world (Gill 1993, Johnston 1996). 
Lesotho is divided into four major ecological zones, namely; the 
lowlands, the foothills, the mountains and the Senqu River Valley. It has 
also four distinct seasons: the spring planting season occurs between 
August and October; the summer months of November to January 
experience the heaviest rainfall; autumn extends from February to April 
and the winter months of May to July bring frost and sometimes much 
snow (Stevens 1967). The country is relatively homogenous in terms of 
linguistic and cultural affiliation. Sesotho is the national language, while 
English language is the medium of instruction in tertiary institutions 
and it is used in government and business transactions. 
Available records indicate that Lesotho has a long history of censustaking 
that dates back to 1875 when a de facto population of 128,000 
persons were counted (Kingdom of Lesotho 1976, 1987). This was 
followed by the 1891 census that recorded a population of 219,000. It is 
remarkable that in the twentieth century, ten population censuses have 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 113
been conducted in Lesotho. Apart from the censuses of 1904,1911 and 
1921, the country has consistently held decennial censuses from 1936 
to date (Ministry of Economic Planning 1996). 
The population of Lesofho increased from 852,000 in 1966 to 1,064,000 
in 1976 and then to 1,447,000 in 1986 and 1,880,146 in 19%. The country's 
population has been estimated to be 2.2 million in 2000 and is projected 
to be 3.2 million by 2020 (United Nations 1999). Thus, in terms of both 
density and size, Lesotho is small. However, the population growth rate 
has not been small. The intercensal (exponential) growth rates for the 
periods 1956-1966,1966-1976,1976-1986 and 1986-1996 have been 
respectively 2.9 per cent, 2.2 per cent, 3.1 per cent and 2.6 per cent. 
World Bank (2000) estimates suggest that life expectancy at birth is 
55 years for Lesotho, while the infant mortality rate is 93 deaths per 
1,000 live births, which is slightly higher than the regional average of 92 
deaths per 1,000 live births. About 49 per cent of the population live 
below the national poverty line, while the gross national product per 
capita is currently US$560. Only 25 per cent of the population of Lesotho 
live in urban areas (United Nations 1998a, Sembajwe 1985). Only about 
17 per cent of the population aged 15 years and over are illiterate, as 
compared with 39 per cent for sub Saharan Africa (World Bank 2000). It 
is remarkable that in Lesotho both literacy and educational levels are 
higher for women than men. This is because more uneducated adult 
men migrate to South Africa to work as unskilled labourers in the 
country's diamond mines (Sembajwe and Makatsjane 1992). Furthermore, 
the predominantly rural cattle-rearers are composed of more males than 
females. 
Data and methods 
The data for this study are from the two most recent censuses of Lesotho. 
These are the 1986 and 1996 census data files that are archived by the 
African Census Analysis Project (ACAP) of the Population Studies 
Center of the University of Pennsylvania. One of the principal ways of 
identifying errors in age data is to examine the single year age 
distributions. In this regard, the single year age distributions of the two 
114 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
censuses will be represented graphically to see whether they conform 
to standard, hi addition, three conventional indices of appraising single 
year age data, namely, the Whipple's index; the Myers' blended index 
and the Bachi's index, will be applied to the two census data sets 
(Shryock and Siegel 1976, Bachi 1951, Myers 1940). The Whipple's 
index is used to measure age preferences for digits ending in 0 and 5 as 
compared with other digits. According to the scheme for estimating the 
reliability of the data, the index varies between a minimum of 100, 
indicating no concentration at all at digits0and5 andamaximum of 500, 
if only digits 0 and 5 were selected. Between these extreme values, the 
data quality is regarded as highly accurate if the Whipple's index is less 
than 105; fairly accurate if the index lies between 105 and 109.9; 
approximate if it falls between 110 and 124.9; rough ifit is located in the 
range of 125—174.9 and, finally, very rough if the index is 175 or more. 
The Myers' blended index is the most widely used test of age 
accuracy for single year distributions. This is because it helps to identify 
whether there is a preference for ages ending in certain digits over 
others in the census enumeration. The index shows the preference for, 
or avoidance of, each of the ten digits between 0 and 9 inclusive. One 
could take the successive sums of numbers which end in each of these 
digits to determine preferences. However, this is not appropriate since, 
with advancing terminal digits of age, the sum would tend to increase. 
The Myers' index circumvents this-problem by blending the population 
in such a way that each digit has almost an equal sum. The blended 
totals for each of the ten digits are expected to be nearly 10 per cent of 
the grand total. The extent, therefore, of the over-selection or avoidance 
of a particular digit shows up in the deviation from 10 per cent of the 
proportion of the total population reporting on the given digit. The 
deviations of each sum from 10 per cent of the grand total are added 
together, ignoring the signs. Their sum is the Myers' index. A summary 
index of preference for all terminal digits is derived as one-half of the 
sum of deviations from 10 per cent. The index is 0 where all terminal 
digits are equally chosen and 90 where a single terminal digit is chosen 
by everyone. 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 115
The Bachi's index is another technique that is used to test accuracy in 
the reported single year-age data. This procedure is similar to the 
Myers' approach. But whereas the Myers' index is based on the sum of 
the absolute deviations of the values, the Bachi's index is based on the 
sum of the positive deviations only. 
~ In assessing the accuracy of reporting by sex and five-year age 
groups, the age ratio score, the sex ratio score and the United Nations 
age-sex accuracy index will be used (United Nations 1952, Shryock and 
Siegel 1976). The United Nations joint score or the age-sex accuracy 
index quantifies the accuracy of the overall age-sex data when the data 
are arranged in five-year age groups. It uses both age and sex ratios to 
identify deviations from what might be expected. The index does not 
only identify digital preferences, but it is also sensitive to the omission 
of changes in the vital rates (Arriaga et al. 1994). The index is computed 
by simply adding the sum of the male age ratio score and the female age 
ratio score to three times the sex ratio score. In defining criteria for data 
quality, the UnitedNations (1983) stipulates that if the index is under 20, 
then data are accurate; if the index lies between 20 and 40, then data are 
inaccurate; and if the index is above 40, then the data are highly 
inaccurate. 
Results 
Patterns of digit preference 
Evidence of distortions in reported census data is clearly shown when 
single year age distributions are considered. Table I indicates patterns 
of digit preference as derived from the 1986 and 19% censuses of 
Lesotho. The table shows that in the 1986 enumeration exercise, digit 0 
was most preferred by both males and females, followed by digits 8,6 
and 2, in mat order. On the other hand, there was a relatively strong 
avoidance of digit 1, followed by digits 9,7 and 3, in mat order. In other 
words, there was a tendency for preference of even numbers and 
avoidance of odd numbers in the 1986 census of Lesotho. However; in 
1996, while digit 6 was the most preferred by males, females preferred 
digit 0 more than any other digit For both sexes, digit 0 was most 
preferred, followed closely by digit 6, while the most avoided digits 
116 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
Table I: Patterns of digit preference, Myers', Bachi's and 
indexes by sex for Lesotho, 1986 and 1996 censuses 
1966 cmsut 
Male Female Both Mais Fs&mate ftofh 
1.2 1,0 11.7) 6 '•"' 2.1 (3.0) 2.2 (3.1) 2.2 (3.0) 
1 -1.9 (-2.8) -1.8 (-2.8) -1.8 (-2.8) -1,0 
(1J) 
H i } 
0.6 OS 
•QX 
1,4{t.l>) 
-1.1 MJ^ -t.l (*1»7) 
{1.4) 
-0.7 tfJB) 
UA <o.$) 0.4 (1.0} 0.4 (0.S) 
-0.2 {-Q-4) -0.1 W<3) 
&m 1.0(1.4} 1.1 (1.7) 
-1.4 H.7) 
1.0 (1-0) 
9.7 
1GS 107 
•2 0.6 (0.4) 0.8 (0.8) 0.7 (0.6) 
3 -0.7 (-0.8) -0.5 (-0.5) -0.6 (-0.6) 
4 0.2 (0.3) 0.3 (0.4) 0.2 (0.3) 
5 0.3 (0.3) 0.3 (0.3) 0.3 (0.3) 
© 1.1 (1.2) 0.9 (1.0) 1.0 (1.1) 1.3 
7 -1.4 (-1.2) -1.5 (-1.2) -1.4 (-1.2) -1.3f«1.8) -1.4 
8 1.2 (1.5) 1.1 (1.3) 1.2 (1.4) 0,9 iQJS) 1.1 (t,0) 
9 -1.5 (-1.6) -1.8 (-2.1) -1.7 (-1.9) -0.9 f1-3) -1,4{-1.8> -1.2 H-€ 
10.9 11.1 11.3 
6.5 6.7 6.8 
115 115 115 
SOURCE: COMPUTED FROM THE 1986 AND 1996 CENSUS DATA FILES OF LESOTHO. 
NOTES 
1. A postth/e sign indicates preference of and a negative sign, avoidance of the digit 
2. Figures in parentheses are results from Bachi method 
3. Myers' index is based on the sum of the absolute values of the deviations; Bachi's 
index on the sum of the positive deviations; Whipple's index on the ages 23-62 
were 7,9 and 1, in that order. On the whole, the patterns of digit preference 
have remained largely the same over the decade. More importantly, the 
magnitude of digit preference and avoidance in Lesotho is very small 
by African standards and highly commendable. 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 117
Further proof of this low digit preference is shown by the computation 
of the evaluative indices employed in the study as revealed by Table I. 
Myers' index declined from 11 in 1986 to 9 in 1996, while Bachi's index 
fell very mildly from 7 to 6 over the decade. Similarly, the Whipple's 
index reduced from 115 in 1986 to 106 in 19%. Going by the assessment 
criteria highlighted in the preceding discussion, these findings reveal 
that there are deficiencies in the Lesotho data. However, the extent of 
age misreporting by sex in Lesotho censuses is not very substantial. 
This is a direct contrast to what obtains in many African populations. 
For example, an assessment of the most recent census age-sex 
distribution of Africa's most populous country, Nigeria, showed that 
the Myers', Bachi's and Whipple's indexes for the nation were 
respectively 62,41 and 293 (National Population Commission 1998). 
The quality of the data has also improved over the decade. 
Sex ratios 
Table n presents the reported age and sex distributions, as well as the 
sex ratios derivable from the 1986 and 19% censuses of Lesotho. 
Lesotho's age structure is characteristic of a young population and is 
typical of a developing country. However, variations in sex ratios (or 
the number of males per 100 females) provide an opportunity for 
assessing the extent of age misreporting in a population by five-year 
age groups. Under normal circumstances, it is a valid expectation in all 
populations to have a slightly higher number of males at young ages 
and a consistently greater number of females at older ages since females 
live longer than males (United Nations 1973, Bicego and Ahmad 19%). 
The table, however, indicates fluctuating sex ratios in both the 1986 and 
1996 censuses. The sex ratio at birth has virtually remained at 101 over 
the decade. But to get a sex ratio of 102 at the age group 35-39 and sex 
ratio of 105 at the age group 45—49, whereas the intervening and 
preceding age groups have lower sex ratios in 1986 is an abnormality 
that point to errors in the age data. Similarly, to have sex ratio of 103 at 
the 45-49 age group in 19% while the preceding age groups have lower 
sex ratios is indicative of data deficiency. In general, the two censuses 
reveal that there are significantly more females than males in the country. 
118 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
1986 Ag© 
Table II: Age and sex distributions and sex ratios from the 
1986 and 1996 censuses of Lesotho 
1996 
group 
Sex ratio Female Male 
101.4 116798 CM 118388 
100.4 114745 5~« 115187 
106271 99.5 10-14105775 
89155 93.1 15-49 82973 
82185 84.5 20-S4 69450 
64032 88.3 25*29' 56551 
51347 93.6 30-34 48036 
102.1 39748 35-39 40590 
33690 99.1 40-44 33373 
29147 105.3 45-49 30681 
31990 93.3 60-54 29851 
20677 95.8 55-58 19814 
17795 88.0 80-64 15666 
18393 82.5 &&-m 15168 
69.7 11801 70-74 8221 
22216 49.8 75+ 11054 
All 
849990 94.2 Ages 800778 
SOURCE: COMPUTED FROM THE 1986 AND 1996 CENSUS DATA FILES OF LESOTHO 
NOTE: Overall, 0.14 per cent in 1986 and 3.60 per cent in 1996 of the population did not 
report their ages. In order to be consistent with the total population size, this proportion 
was distributed evenly (by prorating) across the various age groups on the assumption 
that equal number of "not reported" cases was found in each age group. Sex ratios are 
expressed as male per 100 female. 
The overall sex ratio in Lesotho is 94, which is lower than the model sex 
ratio range of 100-107 for Africa (United Nations 1983). 
The persistence of such a low sex ratio over the decade may be 
attributed to a number of inter-related factors. These include under- 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 119 
Male Female Sex rotto 
105084 101.8 1067S5 
102.1 124406 121883 
101.9 131101 133533 
95,5 121354 115843 
89.0 86977 
92.6 64106 
90.5 56015. 
91.4 47004 
93.9 40166 
102.5 34681 
94.2 27101 
96.4 22165 
79.3 
77.8 
67.2 
97735 
69323 
61863 ' 
51419 
42785 
33845 
28779 
23001 
26686. 
15482 
12938 
23681 
.21169 
12041 
8699 
12367 52.2 
94.4 966989 913187
reporting of females; high male emigration and a low sex ratio at birth, 
modified progressively by high male mortality at subsequent ages. In 
particular, as highlighted in the background information, Lesotho is 
completely surrounded by South Africa and, as a result the migration of 
able-bodied men within the economically productive age category to 
South Africa is common. Therefore, some increase in the sex ratios 
might be expected at older ages (from age 45 and above) if males migrate 
back to Lesotho after having worked in South Africa for some time. 
Age ratios and age accuracy indexes 
Age ratio is computed in this study as the percentage of the population 
in a given five-year age group to the average of the preceding and 
following age groups (Shryock and Siegel 1976). According to the 
scheme of evaluation 100 is the point of balance; age ratios higher than 
100 indicate an over-enumeration of the particular age group, while an 
age group is deemed under-enumerated if its age ratio goes below 100. 
The results presented in the first panel of Table HI show that there are 
distortions in the reported census data of Lesotho. In 1986 overenumeration 
took place at the 5-9,10-14,50-54 and 65-69 age groups 
for both males and females, while the other age groups were underenumerated. 
In 19%, the age groups 5-9,10-14,15-19,30-34,45-49 
and 60-64 were over-enumerated, while the other age groups were underenumerated 
in case of males, At the same time, the age groups 5-9, 
10-14, 15-19, 20-24, 30-34, 40-44, 50-54 and 60-64 were overenumerated 
with respect to females. In alL the largest over-enumeration 
occurred at age group 65-69 for males and 50-54 for females in 1986, 
while the greatest under-enumeration was found in age group 55-59 for 
both sexes. 
Similarly, in 19%, the highest over-enumeration occurred at age group 
60-64 for both sexes, while the largest under-enumeration was found at 
age group 65-69 for both sexes. These findings reveal that the elderly 
persons in Lesotho are more likely to state their ages incorrectly than 
the younger ones, which, no doubt, is linked to poor levels of educational 
attainment on the part of the older population. 
120 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Table III: Age ratios and age-sex accuracy Indexes tor 
Lesotho. 1986 and 1996 
1986 
Male Female Ma!« F»maf« 
NA 
102.8 
106.8 
94.7 
99.6 
963 
98.9 
99.7 
93.7 
97.1 
1182 
87.1 
89.6 
127.0 
NA 
NA 
Uftft 
111.1 
105,1 
NA 
102.9 
104.2 
94.6 
107.3 %£ 
95.9 
99.0 
93.5 
ib&,7 
97.7 
NA 
103.2 
107.8 
106.1 
1025 
86.9 
1025 
9S3 
100.4 983 
9^.6 WBJl 
953 10IJ 
'923 
13$.7 
97.8 
88.8 
128.4 
83.1 
91.1 
9\$ 
W$ 
mi 80.6 
iNfA 
1243 
NA . NA 
30-3* 
1986 Age ratio score for males « 7.6; for females - 9.5 
19S6 Sex ratio score - 6.0; 1986 Age-sex accuracy Index - 35.1 
1996 Age ratio score for males - 7.5; for females - 9.3 
1996 Sex ratio score - 5.1; 1996 Age-sex accuracy index * 32.0 
SOURCE: Computed from the Lesotho 1986 and 1996 census data files. 
NOTE: NA- not applicable. 
The results displayed in the second panel of Table in show that there 
has been a modest improvement in the age-sex accuracy index of 
Lesotho during the study period (from 35 in 1986 to 32 in 19%). Judging 
by the scheme of assessment of data quality, however, the values indicate 
that the Lesotho data are not accurate. But both the 1986 and 1996 
values are significantly lower than what is found in many African 
populations. For instance, the age-sex accuracy index from the 1991 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 121 
Mba/RELIABILITY OF LESOTHO CENSUS DATA
census of Nigeria yielded a value of 81 (National Population Commission 
1998).The corresponding value for the 1983 census of the Gambia was 
85(Makannah 1990). On the other hand, the 1984 census of Tunisia, the 
1981 census of Botswana and the 1979 census of Kenya yielded the 
age-sex accuracy indexes of 25, 26 and 26, respectively (Makannah 
1990), indicating that the quality of these data is good, just as that of 
Lesotho. In fact, in a more recent assessment of the 1991 census of 
Batswana, Udjo (1994) argues that the data are good and concludes 
that the estimates of levels and trends in childhood mortality derived 
from the 1991 census appear to be plausible. 
Discussion and policy issues 
In general, demographic data derivable from age distributions are known 
to have various kinds of errors that ultimately affect the derived 
estimates. The most salient of these errors includes misreporting of the 
dates of birth, or incorrect reporting of the ages of the respondents. 
The present study seems to suggest that the reported age-sex 
distributions of Lesotho's population are fraught with distortions. 
Unfortunately, the true extent and degree of these biases cannot be 
readily determined because of complexities in the patterns of these 
errors. If women, for example, misreported their ages but reported their 
fertility experiences accurately, the estimated fertility rates and implied 
trends in fertility will definitely be biased in some way. 
If, in addition to age misstatement, children ever born were 
underreported or if births were not accurately declared, this will result 
in the confounding of the biases of the fertility estimates. These 
assessment situations could not be addressed in the present study. 
However, within the limits of the evaluative tools employed here, it can 
be argued that the magnitude of the reported distortions is not significant 
enough to warrant the invalidation of demographic estimates based on 
the two data sets of Lesotho. In fact, going by African standards, the 
quality of Lesotho's census data is high and it has improved with the 
passage of time. Moreover, the fact that the results from both censuses 
are similar is also a major consistency check and heightens confidence 
in the reported age-sex distributions of Lesotho. 
122 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
A major contributory factor to the good quality data from Lesotho is 
the influence of education. The latest available empirical evidence 
indicates that* as at 1995, the adult female literacy rate is 92 per cent 
(which is high by world standards), the adult male literacy rate is 69 per 
cent and overall net primary school enrolment for both sexes is 71 per 
cent (World Bank, 2000). Because females are more educated than males, 
they are more likely than males to report their ages accurately. 
Furthermore, the present analysis has shown that the older people are 
more likely to specify their ages incorrectly than the younger ones, a 
condition that has a direct correlation with low levels of education. The 
government should, therefore, do more to encourage more educational 
attainment, especially for males. 
The impact of migration could not be addressed in this study due to 
data availability constraints. It is conceded that migration, as a major 
component of population change, could play a significant role in the 
reported sex-age distributions of Lesotho since available evidence 
shows that there is pronounced out migration of Lesotho indigenes, 
particularly adult males, to neighbouring countries, especially South 
Africa, in search of employment opportunities (Gill 1993, Sembajwe and 
Makatsjane 1992). However, it is doubtful that the volume of emigration 
is the sole contributory factor resulting in the low sex ratios at all ages 
in both censuses. This is because the present analysis has revealed 
that there are considerable low sex ratios at almost all ages in Lesotho. 
On the other hand, if male emigration is negligible, then a low overall 
sex ratio is a product of female under-reporting and/or a low sex ratio at 
birth, modified progressively by high male mortality at all ages since, in 
all human populations, the force of mortality is generally higher for 
males than females (United Nations 1998b, 1991). It should be stated 
that in Lesotho and in other countries of Africa, civil registration and 
vital statistics systems are still developing (Obonyo and Bauni 1999, 
Sembajwe 1990, Moriyama 1982). Asaresult, it is impossible to determine 
the true age distribution of Lesotho without verifiable birth certificates, 
even when migration is taken into account. 
Thus, it. is suggested that the government of Lesotho should pay 
more attention to the issue of registration of births and deaths (the vital 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 123
registration system). This is because the records produced by this 
registration process serve two crucial purposes. In the first instance, 
the individual records establish a person's civil status and the facts on 
which it is based. Proof of these facts such as age, place of birth or 
death, circumstances of death and so on, are essential documents 
required for many official purposes in the society in which a person 
lives or has lived. Secondly, the information on the registration records 
collectively form the data for vital statistics which are the computed 
indices of fertility, mortality and migration, among others, showing 
changes in population size and composition and important features 
regarding the health of a population. Above all, they furnish an empirical 
platform on which decennial census results can be matched and 
assessed. 
References 
Arriaga, E.E. Johnson, P.D. and Jamison, E. 1994. Population Analysis 
with Microcomputers: Presentation ofTechniques. \bl. 7, Washington, 
DC, U.S. Bureau of the Census 
Bachi, R. 1951. The tendency to round off age returns: measurement 
and corrections. Bulletin of the International Statistical Institute, 
Proceedings of the 27th Session, Calcutta, 33 4 195-222 
Bicego, G. and Ahmad, O.B. 1996. Infant and Child Mortality, DHS 
Comparative Studies No. 20. Columbia, Maryland, Institute for Resource 
Development/Macro Systems, Inc. Caldwell, J.C. 1966. A study of age 
misstatement among young children in Ghana 
Demography, 3 2 477-490 
Cleland,J. 1996. Demographic data collection in less developed countries 
1946-19%. Population Studies 50 3 433-450 
El-Attar, M. 1998. Population and rural development: a new approach to 
an old problem with special attention to developing nations. 
Demography India27,1147-153 
Gill, S.J. 1993. A Short History of Lesotho: From the Late Stone Age 
until the 1993 Elections, Morija, Morija Museum and Archives 
124 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL 18 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
Johnston, D. l996.Lesotho, World Bibliographic Series Vol. 3, revised 
edition. Oxford, Clio Press 
Keyfitz, N. and Flieger, W. 1990. World Population Growth and Aging: 
Demographic Trends in the Late Twentieth Century, Chicago and 
London, University of Chicago Press 
Kingdom of Lesotho, 1987. 1986 Population Census: Preliminary 
Results, Maseru, Bureau of Statistics 
— 1966. 7966 Population Census Report Vol. 1. Maseru, Bureau of 
Statistics 
— 1976. 7976 Population Census Report Vol. 77. Maseru, Bureau of 
Statistics 
Krotki, KJ. (ed.), 1978. Developments in Dual System Estimation of 
Population Size and Growth. Edmonton, University of Alberta Press 
Makannah, T.J. 1990. An Evaluation of the Age-Sex Data of Recent 
African Censuses. Paper presented at the Sixth Session of the Joint 
Conference of African Planners, Statisticians and Demographers, Addis 
Ababa, United Nations Economic Commission for Africa 
Ministry of Economic Planning, 19%. 7996 Lesotho Population Census 
Milage List, Maseru, Bureau of Statistics 
Moriyama, I.M. 1982. Potentials of Records and Statistics from Civil 
Registration Systems for Health Administration andTtesea/cfc. Technical 
Paper No. 19. International Institute for Vital Registration and Statistics 
Myers, RJ. 1940. Errors and biases in the reporting of ages in census 
data. Transactions of the Acturial Society of, America 41 395-415 
National Population Commission (NPC), 1998.7997 Population Census 
of the Federal Republic of Nigeria: Analytical Report at the National 
Level. Abuja, Nigeria 
Obonyo, B J. and Bauni, E.K. 1999. Civil registration systems in Africa: 
present and future. Union for African Population Studies, 463-472 
Sembajwe, I. 1985. Urban Population Growth Rates in Africa with 
Special Reference to Lesotho. Working Papers in Demography No. 4, 
Department of Statistics, Demography Unit, Maseru, National University 
of Lesotho 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 125
Sembajwe, I. 1990. The Evaluation of the Vital Registration System in 
Lesotho: The Case of Mantsebo, Working Papers in Demography, 
Department of Statistics, Demography Unit, Maseru, National University 
of Lesotho 
Sembajwe, I. and Makatsjane, T. 1992. Migration and rural crisis in a 
labour reserve economy: Lesotho. In: Toure, M. and Fadayomi, T.O. 
(eds). Migration, Development and Urbanization Policies in Sub- 
Saharan Africa, Council for the Development of Social Science Research 
in Africa, Abijan, 237-276 
Shryock, H.and Siegel, J.S. 1976. The Methods and Materials of 
Demography, (condensed ed. By E.G. Stockwell, Washington, DC, U.S. 
Census Bureau 
Spiegelman, M. 1968. Introduction to Demography, second edition. 
Cambridge, Harvard University Press 
Stevens, R.J. 1967. Lesotho,Botswana and Swaziland: theformerHigh 
Commission Territories in Southern Africa, London, Pall Mall Press 
Swamy, V.S., Saxena, A.K., Palmore, J. A,, Mishra, V, Rele, JJR. and Luther, 
N.Y. 1992. Evaluating the Sample Registration System using Indirect 
Estimates of Fertility and Mortality. Occasional Paper No. 3. New Delhi, 
Office of the Registrar General and Census Commissioner 
Udjo, E.O. 1994. An evaluation of age—sex distribution of Botswana 
from the 1991 Census. Unpublished monograph for Botswana Central 
Statistics Office, Gaborone 
United Nations, 1952. Accuracy tests for census age distributions 
tabulated in five-year and ten-year groups. Population Bulletin. United 
Nations Population Division, No. 2, October 
— 1973. The Determinants and Consequences of Population Trends: 
New Sample of Findings on Interaction of Demographic, Economic 
and Social Factors, vol. 1. New York 
—1983. Manual X: Indirect Techniques for Demographic Estimation. 
New York. Population Division, ST/ESA/SER.A/81 
— 1991. Child Mortality in Developing Countries: Socio-economic 
Differentials, Trends and Implications. New York. Population Division, 
ST/ESA/SERJV123 
126 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA VOL t8 NO 1 JANUARY 2003
Mba/RELIABILITY OF LESOTHO CENSUS DATA 
— 1998a. World Urbanization Prospects. The 1996 Revision. New York. 
Population Division, SlVESA/SER.A/170 
— 1998b. Too Young to Die: Gene or Gender? New York. Population 
Division, ST/ESA/SER.A/155 
— 1999. World Population Prospects. The 1998 Revision: Comprehensive 
Tables. New York, Population Division, ST/ESA/SER.A/177 
World Bank ( 2000). World Development Indicators Data Base, July, 
2000, The World Bank Group 
Acknowledgements 
This research was conducted while Chuks J. Mba (Mbamaonyeukwu, 
C.J.) was a postdoctoral fellow with the African Census Analysis 
Project (ACAP) at the Population Studies Center, University of 
Pennsylvania. Support for this research was provided by the Andrew 
W. Mellon Foundation grant to ACAP. The author wishes to thank 
Prof. Tukufu Zuberifor assistance and the anonymous reviewer for 
useful comments. The views expressed in this article are those of the 
author and do not reflect the official position of the United Nations 
Regional Institute for Population Studies or ACAP 
VOL 18 NO 1 JANUARY 2003 JOURNAL OF SOCIAL DEVELOPMENT IN AFRICA 127