ST18405DCE

UNIT I

Demography: Demography is the scientific study of human population dynamics. It encompasses the
study of size, structure and distribution of population, and how populations change over time due to births,
deaths, migration and ageing. Demographic analysis can relate to whole societies or to groups defined by
criteria such as education, nationality, religion and ethnicity. In most countries demography is regarded
as a branch of either economics or sociology. Formal demography limits its object of study to the
measurement of population process, which is the more broad field of population studies to analyze the
relationship between economic, social, cultural and biological processes influencing the population.
Vital Statistics: Vital Statistics can be defined as the total process of registering, compiling and reporting
of the aggregate of vital events which have to do with an individual’s entrance into life (birth) or exit from
life (death) and change in the social and civil status (marriage, divorce, adoption, etc.) that may occur
during a specified duration of time, among the members of a population residing within a country or any
delimited territory during the same period.
Vital Statistics is defined as that branch of Biometry which deals with data and the laws of human
mortality, morbidity and demography.
Methods of obtaining Vital Statistics: the various methods of obtaining Vital Statistics are given below:
i) Registration method: The most important source of obtaining vital statistics data
is the registration method which consists in continuous and permanent recording of vital events pertaining
to births, deaths, marriages, migration, etc. These, data in addition to their statistical utility also have their
value as legal documents. Registration of births provide information on the place of birth, sex, age and
religion of the parents, legitimacy, number of previous issues& their sexes , father’s occupation and birth
place of parents. Similarly, death registration furnishes information on place of death, sex, age, marital
status, number of issues, birth place, occupation, and cause of death. Similar information is also obtained
with respect to marriages & migrants.
ii) Census Method: Almost in all the countries all over the world population census
is conducted at regular intervals of time, usually ten years. Census consists of complete enumeration of
the population of the particular area under study and collecting information from individual regarding age,
sex, marital status, occupation, religion and other economic and social characteristics. The main drawback
of the census method is that it provides vital statistics only for the census years and fails to give any
information about the vital events in the intercensal period.

1
 ST18405DCE
UNIT I

Population: Population is a group of items, units or subjects which is under reference of study. A
population is constantly in a state of flux. During this process the size of population varies and develops
potentialities for unlimited growth. At any given time t , the growth of population in absolute numbers is
largely determined by its own size at the moment. Thus symbolically, if Pt is the population of defined
territory at time t , then
Pt +Dt = Pt + DPt , the population after the lapse of Dt , will depend upon its initial size Pt & the
duration Dt . If Pt is large, then the increase DPt will also tend to be large in absolute numbers and hence

the chances of Pt + Dt being different from Pt will be great even when Dt is small. On the other hand, if Pt

is very small, then even long interval Dt may lapse without any change being recorded. For empirical
analysis the interval of time usually chosen to study changes in a fairly large population is one year.
Cohort: A cohort is the aggregate of all units that experience a particular demographic event during a
specified time interval. Population can also be sub-divided on the basis of the year of birth of individuals.
All those who are born during a given year form a ‘birth cohort’ or simply the ‘cohort’ of that year. For
example, those who are age 20 in 2005, are the survivors of 1985. Thus, a population at a given time is
the aggregate of the survivors of different cohorts of people born during different years.
Experience shows that different cohorts are not alike in their characteristics as they would have
passed through different social, economic & demographic conditions. For this reason, cohort-wise analysis
of demographic data is necessary to understand the process of changes over time.
The term cohort can be used in respect of other conditions also. A marriage cohort then means all
those, who were married during a given period.
Radix: Rates and ratios are averages expressed as fractions. Death rate, obtained, for example
mathematically, is the average number of deaths per person &is always a value between 0&1. Such values
are always can be confusing and meaningless for ordinary use & give a false sense of accuracy. To avoid
these demerits, rates and ratios are conveniently expressed as per 100 persons or 1000 persons instead of
per person. The multiplier 100, 1000 etc., used to round off decimals is called the radix. Radix could be
thought of as the minimum sample size of the population that would be necessary to obtain a reliable
estimate of a particular rate or ratio. Usually, the radix chosen to describe the sex ratio is 100 & for death
rate it is taken as 1000. For mathematical manipulations the radix is kept as one.
Demographic Data: The major source of information on the demographic process of a country is the
census and the vital statistics registration system. Demographic sample surveys are carried out when these

2
 ST18405DCE
UNIT I

data are not adequate. Secondary data from governmental and other agencies can at times provide useful
information on certain aspects of population not covered by census and vital statistics.
The census is the official survey conducted periodically by the governments to count the people
of its country & collect the information about them. Because of the cost involved & the elaborate legal &
administrative arrangements that extent over several years before and after the actual count, censuses are
conducted only once in 10 years. Therefore, for intervening periods, the size of populations will have to
be estimated.
Statistics on the vital events such as birth, death and marriages are usually maintained by other
agencies of the government. The network of offices all over the country registers the information on these
events, when the event had occurred. The national agency collects these details and publishes annually or
monthly, the consolidated statistics. The total number of events and certain characteristics of the
individuals associated with the events are made available in these reports.
Demographers may at times find that secondary sources such as school enrolment registers,
electoral rolls, and employment/unemployment records, could also provide needed data. When the above
sources are found inadequate and when specific problems have to be studied intensively, special surveys
are carried out.
Demographic Balancing Equation: If birth and death are the two most fundamental demographic
processes, migration is probably the third. The size of the world’s population is (at least at present)
completely determined by birth and death rates, but the population in any particular region or locale is
also determined by the net migration. These three processes are expressed in the demographic balancing
equation – the increase (or decrease) in a population as the algebraic sum of births, deaths, immigration,
and emigration. In general, we can write the difference between two population growths at two time points
0 and t as equal to
Pt - P0 = B - D + I - E (1)

where B, D, I, and E represent the number of births, deaths, immigrations and emigrations in absolute
numbers. Equation (1) is called the balancing equation of population growth and is very fundamental in
concept as it is an equation relating population change to its determinants.
Example: The Demographic Balancing Equation for the United States (From Mcfalls, 1991) (Numbers
in thousands)
Starting Population + (Births-Deaths) + (Immigration-Emigration) = Ending Population

3
 ST18405DCE
UNIT I

Starting Population + (Natural increase) + (Net migration) = Ending Population

World =5,245,071 + (142,959 – 50,418) = 5,245,071 + 92,541 = 5,337,612
U.S. = 248,168 + (4,179 – 2,162) + (853 - 160) = 248,168 + 2,107 + 693= 250,878
Coverage and content errors in demographic data: Demographic data collected by the census as well as
the vital statistics registration cannot be free from biases and errors. The demographer often realizes the
nature and extent of error when he is confronted with the data which are made available to him only in the
form of statistical summary tables. The confidentiality attached to the original records prevents the
possibility of direct checking.
Experience shows that there will be more errors than are often believed. Errors could occur at any
stage: planning, organizing, recording, compiling, tabulating, analyzing or publishing. Defective
schedules, enumerator bias, response errors, administrative lapses and clerical mistakes are some of the
errors one can expect. There can be no definite test that could be conclusive but, certain tests could provide
sufficient insight on the quality of the data. A direct way to check the total is to investigate the growth of
population during the intercensal period. For this, balancing equation could be used properly when fairly
accurate vital statistics and migration data are available. Then using the balancing equation
Pt - P0 = B - D + I - E
where Pt and P0 are populations at two time points 0& t, and B, D, I, and E represents the number of births,
deaths, immigrations and emigrations in absolute numbers. We have the difference between the right and
left expressions of the balancing equation as the error of coverage.
Coverage error = ( Pt - P0 ) - ( B - D + I - E ) (1)

The expression on the right side of (1) is known as error of closure in a different context, i.e. when the
vital statistics or migration data are not completely reliable.
If rates are more dependable then growth of the period arrived at by comparing population totals
could be checked with the growth rates calculated using vital rates. Thus if

1 æP ö
r= logçç t ÷÷
t è P0 ø
is the rate of growth base upon population totals and
r¢ = b - d + i - e
is the average rate of the intercensal period based on birth, death and migration rates, then, when counts
are correct, r and r ¢ will be close to each other.

4
 ST18405DCE
UNIT I

Chandrasekharan-Deming Formula: Several kinds of checks are available to verify the accuracy of
published reports. The method suggested by Chandrasekharan and Deming to evaluate the coverage of
vital registration and to estimate the undercount is different from the conventional approaches. The method
is described below: -
The Chandrasekharan-Deming method envisages a second agency collecting on vital events for a
pre-assigned period. The second agency could even be a retrospective survey. The two sets of data are
matched event and the results could be categorized into a 2 ´ 2 table as follows:
Registration
Registered Not-registered Total
Information A C A+C
Independent gathered
Agency Could not B D B+D
detect
Total A+B C+D N=A+B+C+D

Here A is the number of events that were registered and was also gathered by the independent agency. B
is the number of events that were registered but failed to be detected by the agency. C is the number of
events detected by the agency which went unregistered. Thus, (A+B) is the total number of events
registered and (A+C) is the total number of events collected by the agency. Naturally, there is the
possibility that a few events could have been missed by both the Registration & the Agency, and D is that
unknown number. The problem is of estimating D and once D is known, the amount of undercount could
be estimated.
If constant proportion of undercount is assumed for both agencies, or in statistical terminology, if
the chance of detecting an event is independent of not detecting that event for both agencies, then we have
A C A+C
= =
A+ B C + D N
On rearranging and simplifying, we can get
BC
Dˆ =
A
where D̂ is the estimate of D. as the right side of the above equation is known, D is easily estimated. This
simple formula of Chandrasekharan and Deming has turned ot to be very useful in evaluating vital
5
 ST18405DCE
UNIT I

registration in many cases when it is suspected to be inferior. The Chandrasekharan-Deming formula is

valid only the assumption of independence holds good.
Age: Age of an individual is the interval of time from birth to time of ascertaining. The definition of age
differs among cultures and under different situations. For demographic purposes, it is agreed that the
number of full years lived by an individual, ignoring fraction thereof.
Accuracy of Data on age: Age although the most important demographic variable is rarely free from
reporting bias. Age is often deliberately misreported for petty reasons. Age data have been found to be
subject to tendentious errors, particularly age heaping. The tendency for people to report an age closer to
their actual age because of certain importance or emotional reasons attached to it is called tendentious bias.
The tendency of a person near 16 to report the age as 18 because 18 are significant for voting rights, is an
example. Yet another strong tendency among persons while reporting there is to exhibit digit preference,
that is, unexplained liking for numbers ending in certain digits like 0 and 5. The result of such preference
is that there will be unusually large number of persons at ages ending with preferred digit. This is known
as age heaping. A single year age distribution suffering from age heaping will look like a zigzag uneven
curve, with peaks at certain ages ending in 0, 5, etc. and unusual depressions at the preceding succeeding
digits like 9 & 1, when ‘0’ is preferred and 4 & 6 when ‘5’ is preferred. Most of the census age returns
suffer from this bias. Because of these biases a demographer has to wary about the tendentious bias,
particularly age heaping.
Some of the well known methods of evaluating age data:
Whipple’s Index: Whipple’s index gives the relative preference of the digits ‘0’ & ‘5’ while reporting age
in the age interval 23 to 62. It is taken as
P25 + P30 + ... + P60
W .I = ´ 100
1 62
å Px
5 x = 23
where Px is the number of persons reporting their age as x. from the above expression it is clear that if

there is no universal tendency to report ages as 25, 30, 35, etc., the numerator should be approximately
1/5th of the total population in the range of 23 to 62 years given in the denominator, because ‘0’ and ‘5’
are just 2 out of 10 digits 0, 1, …, 9. Thus if there is no age heaping at 0 and 5, then the index have a value
of 100, as the theoretical minimum. The other extreme is obtained when there is complete age heaping;
that is, when the entire population is concentrated at ages 25, 30… etc. with none reporting their ages as
23, 24, 26, … , 29,31, …etc; in which case the index will have a value of 500.
6
 ST18405DCE
UNIT I

Studies of age have shown that preference for age is not limited to 0 & 5 but to some other digits
also but with no particular pattern. In such situations the index as defined by Whipple is not helpful.
Further, the index considers only the arbitrary interval 23 to 62 years and not the entire life span. The most
serious drawback however, is that Whipple’s index does not take into consideration the decreasing nature
of the age distribution due to depletion by death. In other words population of younger ages would receive
greater weight-age than at the older ages which would bias the index. To over these drawbacks Myer has
suggested a modified index.
Myer’s Index: Myer’s approach is to derive a ‘blended population’ to overcome the decreasing nature of
the population curve and to calculate heaping at each age using the entire life span.
Blended population
The blended population is calculated as follows. The first is to write the population in a matrix
form and sum the rows as given below:
Population Sum of each row
at each age From 1 to 9 From 2 to 9
P10 P20 … P90 X0 Y0
P11 P21 … P91 X1 Y1
P12 P22 … P92 X2 Y2
…………………………..... ……….. ………..

……………………………. ……….. ………..

P19 P29 … P99 X9 Y9

Here Pij is the population reporting the age as (i j) where each row corresponds to j= 0, 1, 2… 9 or the
“preferred” digit. The sum for digit is obtained for intervals (10-99) and then for (20-99) which are denoted
by Xj’s and Yj’s using weights.
Xj X0 X1 X2 X3 X4 X5 X6 X7 X8 X9
Weights 1 2 3 4 5 6 7 8 9 10
Yj Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9
Weights 9 8 7 6 5 4 3 2 1 0
So that we have each of the preferred digits the corresponding “blended population”.
Preferred digits Weighted Blended
sums population
0 1X0 +9Y0 B0

7
 ST18405DCE
UNIT I

1 2X1 +8Y1 B1
2 3X2 +7Y2 B2
… … ………..

… …. ………..

9 10X9 +0Y9 B9

9
Note that the total weights are same for each digit. Any Bj must be 10% of åB
j =0
j when there is

no preference for the digit j. hence the index can be defined as

M .I = å| b j - 10 |

Bj
Where bj = 9
´ 100
åB
j =0
j

When there is no absolute age heaping at any digit each b j will be 10 and so the index will be equal to

zero. The extreme is when there is age heaping at one digit with all other digits being not preferred in
which case Myer’s index will be take the value 180.
Myer’s index too is not above criticism. Like the Whipple’s index this one too does not have any
sound theoretical basis other than perhaps, rectilinearity of the age distribution which as we know is not
very realistic. And digit preference is not the only tendentious bias in reported age distribution, and this
index cannot assess the other biases.
There are several other measures like Baachi’s index and Ramachandran’s index to measure age
heaping. But Myer’s index is the simplest among them and is still in common use particularly because
other indices have not established their superiority.
Example: Using age distribution data for a West African country (1960), determine the extent of heaping
on
(a) Digit ‘0’ and ‘5’.
(b) Digit ‘0’ only
(c) Digit ‘5’ only.
Age Population Age Population
23 38687 45 44654

8
 ST18405DCE
UNIT I

24 51289 46-49 77102

25 77141 50 52024
26-29 201450 51-54 44751
30 110379 55 14024
31-34 132136 56-59 45283
35 64091 60 38377
36-39 134140 61 4779
40 81515 62 7866
41-44 84422 Total 1304110
Solution:
P25 + P30 + ... + P60
W .I on digit 0 and 5 = ´ 100
1 62
å Px
5 x =23
77141 + 110379 + 64091 + 81515 + 44654 + 52024 + 14024 + 38377
= ´ 100
1304110
5
482205
= ´ 100 = 1.849 ´ 100 = 184.9
260822
P30 + .P40 .. + P60
W .I on digit 0 = ´ 100
1 62
å Px
10 x =23
110379 + 81515 + 52024 + 38377
= ´ 100
1304110
10
282295
= ´ 100 = 2.165 ´ 100 = 216.5
130411
P25 + P35 + ... + P55
W .I on digit 5 = ´ 100
1 62
å Px
10 x =23
77141 + 64091 + 44654 + 14024 199910
= ´ 100 = ´ 100 = 1.533 ´ 100 = 153.3
1304110 130411
10

Example: Use the Myer’s blended index to assess the quality of age data given below-

9
 ST18405DCE
UNIT I

Digit 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99
0 386 417 307 190 89 80 50 40 10
1 133 93 90 89 41 35 33 10 2
2 341 227 100 40 30 25 15 6 4
3 223 160 90 46 28 22 8 5 1
4 201 138 50 38 25 10 8 5 3
5 298 238 201 154 148 65 25 2 3
6 198 105 70 49 31 18 12 6 4
7 166 86 50 45 25 9 5 4 2
8 255 100 90 40 28 10 5 5 2
9 132 85 40 20 8 6 2 2 2

Procedure for computations

1. Sum all the populations ending in each digit over the whole range i.e. 10-99
2. Sum figures between ages 20-99.
3. Multiply the sums in (1) by weights; 1, 2, 3,4,5,6,7,8,9 and 10.
4. Multiply the sums in (2) by weights from 9 descending to 0 i.e. 9 ,8 ,7, 6 , 5 ,4 ,3 ,2 ,1 ,0.
5. Add the product of (3) and (4), to obtain the blended sum
6. Add up the blended sum in (5).
7. Find the percent (%) of the total blended sum at different digit ends.
8. Take the deviations of each % in (7) from 10.0. This result indicates the extent of concentration or
avoidance of a particular digit.

Solution:
Dig Sum Weight Product Sum Weight Produc Blende Percen Deviation Remarks
it (10- s (20-99) s t d sum t Dist from 10%
99) bj bj-10
0 1569 1 1569 1183 9 10647 12216 21.7 11.7 Preference
1 526 2 1052 393 8 3144 4196 7.5 -2.5 Avoidance
2 788 3 2364 447 7 3129 5493 9.8 -0.2 Avoidance

10
 ST18405DCE
UNIT I

3 583 4 2332 360 6 2160 4492 8.0 -2.0 Avoidance

4 478 5 2390 277 5 1385 3775 6.7 -3.3 Avoidance
5 1134 6 6804 836 4 3344 10148 18.0 8.0 Preference
6 493 7 3451 295 3 885 4336 7.7 -2.3 Avoidance
7 392 8 3136 226 2 452 3588 6.4 -3.6 Avoidance
8 535 9 4815 280 1 280 5095 9.0 -1.0 Avoidance
9 297 10 2970 165 0 0 2970 5.3 -4.7 Avoidance
Total 56309

M .I = å | b j - 10 |= 39.3

Dependency Ratio: A measure of the portion of a population which is composed of dependents (people
who are too young or too old to work). The dependency ratio is equal to the number of individuals aged
below 15 or above 64 divided by the number of individuals aged 15 to 64, expressed as a percentage. A
rising dependency ratio is a concern in many countries that are facing an aging population, since it
becomes difficult for pension and social security systems to provide for a significantly older, non-working
population.

Population below15 & above 64 years

Dependency ratio = ´ 100
Population between 15 - 64 years

A worked example should make this clearer. Pakistan, which is a developing country, has 41% of its
population less than 15 years and 4% above 64 years, the dependency ratio is given by
41 + 4 45
Dependency ratio = ´ 100 = ´ 100 = 81.8%
100 - (41 + 4) 55
New Zealand, a developed country, has23% of its population less than 15 years and 12% above 64 years,
the dependency ratio is given by
23 + 12 35
Dependency ratio = ´ 100 = ´ 100 = 53.8%
100 - (23 + 12) 65
Countries that have a high dependency ratio have more people who are not of working age and few who
are working and paying taxes. The higher the number, the more people need looking after.

11