Chapter 2 - Evaluation of Data
Evaluation
or appraisal means the measurement of achievement against goals.
Evaluation techniques are often necessary to determine whether the data
collection was properly done and that the data are of acceptable
quality. A
census, being a massive data collection exercise involving thousands of
field interviewers, errors may creep in at any stage of data collection
and processing. An evaluation of the census data is desirable to assess
the quality the data. Evaluative studies probe into the qualitative and
the quantitative aspect of the data. Errors being probed fall into two
broad headings - coverage and content errors. According to the United
Nations, good census practice requires a careful consideration, and an
evaluation of the completeness and accuracy of census results. The
three main objectives of the evaluation are: (i)
To identify the types and sources of errors or biases in order to
know which groups, items or methodology produce the errors. (ii)
To measure the accuracy of the data. (iii)
To adjust the data by taking into account the varieties and
amount of errors present. Broadly,
two methods are available for census evaluation, the direct and the
indirect method. The direct method involves the comparison of
information collected in a census with data from other sources such as
post-enumeration sample surveys or re-enumeration. This method has not
been used for the 2000 census evaluation. Instead, the indirect method
consisting of external and internal consistency checks has been used. 2.2.1 Introduction Coverage
error is a non-sampling error that may occur in various forms, namely: ·
Omitting a unit that should have been included. ·
Including a unit more than once ·
Including a unit that should not have been included. In
spite of the fact that massive efforts were made in providing good
training and up-to-date cartographic maps to the field staff as well as
introducing checks and controls throughout the fieldwork, such errors
may still occur. 2.2.2 Balancing equation by sex The
balancing equation is one of the methods used for detecting coverage
errors. Table
2.1 compares the 2000 enumerated population of the Republic of
Mauritius, with the expected population based on the previous census,
vital statistics and international migration data. Table
2.1-Balancing equation by sex, 2000 Census - Republic of Mauritius
The
table indicates a net deficit of 4,925 persons (-0.42%), that is a
deficit of 6,896 males (-1.18%) and an excess of 1,971 among females
(0.33%). Similar calculations for the 1990 census showed a net deficit
of 474 (-0.04%) with an excess of 719 for males (0.14%) and a deficit of
1,193 (-0.23%) for females. Though the coverage errors seem to be more
pronounced in the 2000 census, they are acceptable when compared to
international norms. An
analysis of inter-censal growth rates can also help in the evaluation of
the census data. Table 2.2 gives the average annual growth rate for the
last four inter-censal periods. Table
2.2-Population annual growth rates (%), 1962-2000 - Republic of
Mauritius
The
natural growth rate shows a smooth decline over time depicting
essentially a fall in fertility registered over the years. However, the
same smooth decline is not observed with net growth rate. This is due to
the fact that during the period 1983-1990 significant increase in
out-migration, especially among females was registered and hence a
relatively low growth rate. The
net growth rate when compared by sex, generally indicates higher values
for females; this is essentially due to lower mortality among females.
However, despite this, the situation was different in 1983-90 mainly due
to more out-migration of females. The
distribution of the population by sex and broad age groups for the last
four censuses is shown in Table 2.3. Table
2.3 - Age composition (%) of the population by sex: 1972, 1983, 1990 and
2000 Censuses - Republic of Mauritius
The
change in the age distribution from one census to another can be
summarised as follows: ·
A decrease in the proportion of children aged 0-4 and 5-14 over
time ·
An increase in the proportion aged 15-44 up to 1990 followed by a
slight fall in 2000 ·
A general decline in the proportion of males aged 45-59 followed
by a sudden increase in 2000 while among females, a general increase is
observed ·
A general increase in the proportion aged 60 years and above The
decline in the proportion aged 0-4 and 5-14 is mainly due to a general
fall in fertility registered over the years.
The jump in the proportion of the population aged 15-44 from
about 43% in 1972 to 50% in 1983 is due to births of the high fertility
period 1957-1968 entering that age group. Average number of births
during that period was around 26,000 annually. Birth cohorts leaving
that age group came from relatively low fertility years (1927-1938).
Similarly increases registered in 1990 are due to birth cohorts 1968-75
entering that age group, during which 21,500 births were registered
annually. The slight fall registered in 2000 could be due to the fact
that the disparity in the size of birth cohorts entering and leaving
that age group was only slight. The
large increase in the proportion aged 45-59 in 2000 is mainly due to the
entrance of post-war baby boomers (born during period 1945-55) into that
age group. The tendency towards increasing proportion aged 60+ with time
is an indication of an ageing population. The higher proportion of
females among the elderly is the consequence of lower mortality among
females. Table
2.4 - Mean and median age of the population - Republic of Mauritius
The
general rise in the mean and median age of the population is also
indicative of the process of ageing under way in the population. Both
the mean and the median have increased by around 3 years during the
intercensal period 1990-2000. It is also noted that the mean and median
age are higher for females than for males. This is again due to
differential mortality between males and females whereby females live
longer than males. The
sex ratio is defined as the number of males per 100 females. The table
below shows a general fall in sex ratio except for the period 1983-1990
when a slight increase was registered. Table
2.5 - Sex ratio of the population - Republic of Mauritius
The
trend in sex ratio over the years is influenced by the degree of sex
differentials in the three factors affecting the change in population
size and structure, namely fertility, mortality and migration. The sex
ratio at birth which is usually above 100, indicating more males being
born than females, tends to increase the overall sex ratio of the
population. Net migration can either increase or decrease the overall
sex ratio while a differential in mortality in favour of females
depresses the overall sex ratio. The mortality factor usually being the
most influential, tends to depress overall sex ratio over time. The
different situation noted during the period 1983-1990 was mainly due to
higher out-migration of females compared to males. The
child-woman ratio is defined as the number of children aged 0-4 years
per 1,000 women in the age group 15-44 years. It is a crude measure of
the level of fertility derived from census data. Table 2.6 shows a
general decline in the child-woman ratio from 473.7 in 1983 to 363.3 in
1990 and 316.7 in 2000 for the Republic. It declined both in the Island
of Mauritius and the Island of Rodrigues over the same period, though
the decline in the latter island was sharper. It should however be noted
that in spite of the sharper fall in fertility in Rodrigues, the level
of the child-woman ratio is still higher than in the Island of
Mauritius. The
above observations are in line with the fertility decline registered
over the past two decades. In fact, the TFR (defined as the average
number of children born to an average woman assuming that she survives
to the end of her child-bearing age and is subject to a fixed schedule
of age-specific fertility rate) in the Island of Mauritius fell from
2.29 in 1990 to 1.99 in 2000, that is by about 13.1%, while in the
Island of Rodrigues it fell from 3.19 in 1990 to 2.70 in 2000, that is
by about 15.4%. Table
2.6 - Child-woman ratio at the 1983, 1990 and 2000 Censuses - Republic
of Mauritius
The
dependency ratio represents the ratio of the combined child population
(0-14 years) and the aged population (65+ years) to the population of
intermediate age (15-64 years). It is a rough measure of economic burden
the productive population has to bear. It can be split into child
dependency and old age dependency. The
table below shows the dependency ratios calculated for the last three
censuses. The figures indicate a general decline in the total dependency
ratio from 588 in 1983 to 539 in 1990 and 460 in 2000. This fall is
mainly attributable to a fall in child dependency brought about by a
fertility decline. It has also been observed that there has been a
continuous increase in the old-age dependency ratio. This is the result
of both fertility decline and mortality improvement. Table
2.7 - Dependency ratio by sex: 1983, 1990 and 2000 Censuses - Republic
of Mauritius
Another
salient feature visible in the data is the higher old age dependency
among females than among males again due to the fact that women live
longer than men. 2.3
External consistency checks 2.3.1 Comparison of Population Census with Housing Census
count The
Housing Census was conducted from February to April 2000 while the
Population Census was taken at the beginning of July 2000. At the
Housing Census, 297,881 private households and 1,168,495 persons were
enumerated compared to 296,832 private households and 1,165,570 persons
at the Population Census. Thus a minor differences of around 0.3% in the
population figures and 0.4% in the household figures are observed. Given
that the Housing Census enumeration covered a period of 3 months
(February-April 2000), it may happen that some households who own
secondary residences have been counted at both their principal and
secondary residences at the Housing Census. However this was not the
case with the Population Census, which was taken on a specific census
date (night of 2-3 July 2000). 2.3.2 Comparison of Population Census with education
statistics Education
statistics is yet another external source of data with which Census data
can be compared. Figures from the 2000 survey conducted in March by the
Ministry of Education are compared with census data on students
currently going to school in table 2.8 below. Table
2.8 - Comparison of 2000 Census data on school population by age group
and sex with statistics from the school system - Republic of Mauritius
According
to the table, deficits of 6.2% among males and 6.0% among females in the
age group 5-9 years are noted in the census figures, indicating that
there may be some under-enumeration. In the age group 15-19 years, an
over-enumeration of about 11% is noted in the census data for both males
and females. It should be pointed out that school statistics as regards
enrolment in vocational or post-secondary schools are not complete. On
the whole however, there is an over- reporting of school attendance at
the census of the order of 0.9% and 0.6% among male and females
respectively. The
discrepancy being relatively small, it can be concluded that there is
compatibility between the two sources of data. 2.3.3 Comparison of Census data with population
estimates. The
expected population for year 2000 was made by surviving the adjusted
1990 Census figures by age and sex using: (i)
Live births data by sex (ii)
Deaths by age and sex (iii)
International migration data by age and sex Since
data for international migration are not available for Rodrigues,
comparison will be restricted to the Island of Mauritius only. Tables
2.9 (a) and 2.9(b) compare the enumerated resident population with the
expected population. The tables show that discrepancies occur mostly in
the age bracket 0-9 years and 20-39 years among both males and females.
Discrepancies occurring at age 0-9 years are attributable to
under-enumeration of young children. Differences
occurring in the age bracket 20-39 years are negative among males and
positive among females. This is mainly attributable to the poor quality
of passenger traffic data as regards sex ratio of migrants. The sex
ratio used in calculating intercensal population estimates was in favour
of females, that is there were more females migrating, while census
results indicate the contrary. This
also leads to population estimates with a higher sex ratio than census
figures. Some degree of age mis-reporting may also have contributed to
the discrepancies observed.
|
