Data as a
SAID/ R. & DJ W.I.D.
The GENESYS Project
I'm going to speak with you today about using demographic data
as a starting point for performing gender analysis. Specifically, I'm
going to demonstrate how sex-disaggregated data--readily
available from most libraries--can be used by development
practitioners to identify gender imbalances that might have
significant implications for the shape and success of development
programs and projects.
Once you have figured out what kinds of data you want to look at,
you then need to know where to go to find these data. Almost all
of the data shown in this presentation were taken from one or
more of the sources shown on this slide. These sources are readily
accessible in most central or university libraries. One can also
contact data sources directly and ask for specific types of data
on specific countries. Various electronic databases containing
international demographic and economic data are also accessible
for modest subscription fees. W.I.D./GENESYS staff are also
available for assistance in procuring relevant data for any specific
I. Concepts and Definitions
Sex and gender
Quantitative and qualitative analysis
II. Examples of Uses of Quantitative
Data in Gender Analysis
imbalanced sex ratios and rural/urban
gender differences in life expectancy
gender differences in education
gender differences in economic activity
incorporating gender in family planning
This is a general outline of what the discussion will look like.
First, we need to clarify a few concepts...
In everyday language, most people think that "sex" and "gender" are
synonymous, and if there is any difference at all, it is that the
term, gender, is the more "politically correct" these days.
Sex differences between males and females are, strictly speaking,
biological in nature. As such they are essentially unchangeable and
universal throughout all human societies.
Gender differences, on the other hand, are cultural in nature. That
is, they are learned behaviors and preferences that are typically
associated with one sex or the other within a given culture. As
such, gender differences are both variable between cultures and are
changeable over time
M Sex differences have been used
as criteria for gendered division of
Often, people have difficulty in distinguishing between sex and
gender because the two concepts have been so intricately
intertwined. For example, sex differences in child bearing and in
physiological strength have often been used as criteria for
determining the division of labor and also the distribution of
resources in many societies. While it is important to note that
advances in modern technology have made sex differences
between men and women less relevant in determining their roles
and relationships, it is also important to emphasize that sex
differences have rarely been absolute determinants of the
division of labor in any society. In other words, most of the
differences in men's and women's roles in every society are due
more to cultural traditions that have evolved over time than
they are to biological characteristics that dictate certain
kinds of labor for women and other kinds for men.
As the name implies, Gender Analysis is primarily concerned with
the cultural dimensions of men's and women's roles and
relationships, and particularly with how these highly variable
cultural characteristics interact with the processes of sustainable
Gender Analysis and
In terms of the relationships between gender differences and
sustainable development, Gender Analysis is, of course,
interested in questions of efficiency and also of equity in
gendered divisions of labor and resource distributions. But, the
principle concern of Gender Analysis is to understand the
social reality of a given developmental situation so that
intervention strategies can be custom designed and
implemented for optimum effectiveness within a particular
The first benefit of Gender Analysis, therefore, is to assist in
the targeting of resources, beneficiaries and activities
according to the social and cultural realities of particular
"successfully anticipate the impacts that interventions will have.
Quantitative data analysis--which is the focue of this presentation
-- can be a useful starting point for identifying potential gender
issues in development that warrant further study...
It is important to point out, however, that while quantitative data
may be necessary, it is not sufficient by itself for performing
Gender Analysis. Quantitative analysis is particularly useful for
describing what is happening and to whom it is happening (or not
Quantitative Analysis can tell us...
* To whom?
Qualitative Analysis, on the other hand, is needed for the
interpretation of quantitative results. Qualitative Analysis
provides valuable insights into How something is happening and
Why it is Happening.
Further, Qualitative analysis is necessary to provide the
meanings by which relevant quantitative research is designed
Qualitative Analysis explains...
How things are
Why are they
Neither type of analysis is sufficient in itself and neither type
can be of much value independent of the other. The task of the
gender analysts is to insightfully integrate the knowledge
gained from both qualitative and quantitative methods into a
comprehensive picture of social reality of a given situation.
Armed with such knowledge, development practitioners can
design their interventions with greater confidence that they will
have significant desired impacts on their target populations.
Examples of using
quantitative data to:
identify gender issues in
formulate relevant questions
for gender analysis.
Now, we are going to look at some examples of using
Quantitative data to:
Identify potential gender issues in development
begin formulating the right kinds of Questions for
It is important that we emphasize that the kinds of data that
we are looking at today are not part of a comprehensive social
analysis, but rather provide a means for focusing our attention
on situations that warrant more indepth analysis.
Sex Ratio Imbalances & Rural Urban
Gender Differentials in Life Expedtancy
Gender Differentials in Educational
Attainment & Economic Activity
Gender Considerations in Unmet Need for
First we are going to examine some sex ratio imbalances and
their implications for a variety of developmental issues. In this
section of the presentation we will also illustrate the impor-
tance of looking at rural and urban differences in gender issues.
Secondly, we look at certain dimensions of gender differences in
life expectancies as indicators of possible gender issues.
We also look at differences in educational attainment and
economic activity by gender.
Lastly, we consider the incorporation of the gender variable into
analysis of unmet need for family planning services using data
from Demographic and Health Surveys.
The first area we can look for potential gender issues is in the
basic age and sex structure of a population. This chart should look
familiar to many of you. It is a population pyramid, or more
precisely, two population pyramids, for Brazil in 1940 &80,
superimposed on each other.
This mode of presentation is useful for illustrating the conse-
quences of rapid population growth, but it is not an easy or precise
way to look for imbalances in sex ratios, which is the subject of our
It may not be readily apparent to you that sex ratio imbalances
can be useful indicators of potential gender issues, but they can.
Overall, the sex ratio of a pop--expressed as the # of males for
every 100 females--are usually fairly close to being balanced. That
is, there are about as many men as women in the overall
population, but there are typically more men in the younger ages
and more women in the most senior years. When we look at an age-
specific distribution of men and women and find that there are
large shortages (or surpluses) of either men or womenin some or all
age-categories we suspect that something is happening in this
society that is a potential mender issue.
We know that there can only be two causes of large sex ratio
 gender differentials in mortality meaning that one gender is at
greater risk of dying than the other, or [
 gender differentials in migration patterns where more of one
gender more than the other is migrating either in or out of a
Either of these causes raise potential gender issues, but of
Here we see a line graph of the age-specific sex ratios of Burkina
Faso in 1985. The horizontal white line indicates the point at which
there would be an equal number of men and women in each age
category. Note that there are large shortages of males partic-
ularly in the prime productive ages from 20 to 60. While these
imbalances are more pronounced in Burkina than elsewhere, this is
a rather typical pattern for several countries in sub-saharan
Africa and of other regions where there are high levels of
subsistence agriculture, rapid population growth and increasing
strains on the carrying capacities of local ecosystems. In these
situations, there are strong pushes for out migration in search of
wage labor in more productive regions. In some places--Latin
America and the Philippines, for example--this migration tends to
be female. In Africa and in most of Asia, however, the out-
migration is more male dominated.
Age-Specific Sex Ratios of
Burkina Faso, 1985
Most of the economically motivated migration today is from
rural to urban areas and a large portion of this migration
takes place within national boundaries.
In Burkina Faso, however, most of the migration is across
national boundaries as can be deduced from this graph which
shows the age- and sex-specific rural and urban populations of
Burkina in proportion to each other.
First, you can see that the rural population is much larger
than the rural. Furthermore, a closer look reveals that while
there are slight surpluses of males in some age categories of
the urban population, they are nowhere near large enough to
account for the shortages of men seen in the rural side of the
This observation tells us that it is predominantly rural males
between the ages of 20 and 60 that are migrating out of
Burkina in large numbers. This, in turn, suggests that any
developmental problems that might be related to the sex
ratio imbalances are more pertinent to the rural than to the
urban population of Burkina.
Implications of Male
Migration Patterns in
SAre agricultural and environmental
programs targeted toward women?
What resources do these women need to
fulfill their new economic and social roles.
Are there legal barriers for women that need
to be changed in this situation?
These sex ratio imbalances in Burkina potentially raise
questions about developmental intervention strategies in all
of the sectoral areas of USAID policies and programs.
Agricultural assistance and environmental protection
projects, for example, could benefit from the knowledge that
women are carrying a greater share of the village work load in
the men's absence. Programs to promote democratization
might also want to focus on issues of women's empowerment
where these issues may affect the productivity of Burkina's
Some Implications of
Male Migration Patterns
for Population & Health
Programs in Burkina Faso
How do the migration patterns of men
influence the spacing of births?
How do the extra labor demands placed on
women due to husbands' absence influence
women's desire for more children?
What are husbands' attitudes about their wives
use of contraceptives in their absence?
This gender-specific migration pattern also has implications
for family planning and health care strategies...Concerning
these, here are a couple of questions that might arise from
observing the migration patterns...[Read Slide]
Here are some more...
Some Implications of
Male Migration Patterns
for Population & Health
Programs in Burkina Faso
* What are the patterns of sexual behavior of the
husbands when they are away from home?
*What are the consequences of these behaviors
for the spread of sexually transmitted diseases
back to the rural villages?
"the average number of
years that a person can
expect to live beyond a
certain age given the current
mortality risks of that
We turn now to an illustration of the use of life expectancy
rates for identifying potential gender issues. Life
expectancy at birth is the most common rate used and is
defined as the average number of years that a person can
expect to live given the current mortality risks of that
We should point out that, by virtue of the way life expectancy
rates are calculated, they are strongly influenced by infant
and child deaths. A society may have a life expectancy at
birth of only 60 years, but if a member of that society sur-
vives past the age of 5, he or she may have an life expec-
tancy of 70 or more years.
Life Expectancy and Living Conditions
Life Expectancy at Birth in the 25 Richest
and 43 Poorest Countries
I Source: World Development Report, 1992
3 Taking these qualifications into account, Life Expc'y rates can have several
general analytical values. Often they are used as a crude indicator of general
standard of living of a population. In general, the higher the level of socio-
economic development of a society, the lower the mortality risks.
For example, this chart shows the average life expectancies of men and women
for the 25 riches countries and the 43 poorest countries in the world. In the
richest countries people average almost three decades more longevity than in
the poorest countries.
Examination of differentials in life expectancy rates can suggest some
important things about the equitable and efficient distributions of resources
and about the divisions of labor in a society. These differentials may also have
several potential implications for the causes of population growth.
Life Expectancy of Women and Men
EXTRA! EXTRA! READ ALL ABOUT IT.
Women outlive men in 181
out of 187 nations!!!
r.y 10% in the
Richest, 4% In
4. The bulk of mortality data suggest that when environmental
factors are held constant, women tend to survive better than
men. Worldwide, there are approximately 105 male babies born for
every 100 female babies, but the sex ratio tends to even out by
the midyears of life and there are typically more women than men
among the very old.
If there is a biological dimension contributing to life expec-
tancy differentials of men and women, it is impossible to
determine the magnitude of this dimension because cultural
and environmental factors strongly affect life expectancy
rates and produce a great deal of variation across societies..
We conclude, therefore, that gender differentials in life
expectancies involve both sex and gender differences. That is,
they are due to a combination of biological and cultural
Life Expectancy and Living Conditions
Average Life Expectancy at Birth in the 25 Richest
and 43 Poorest Countries
70 a Females
Source: World Development Report, 1992
b. Looking again at our comparison of life expectancy rates in
the richest and poorest countries, we note that, on average
Women have about 10% greater longevity among the 25
richest countries of the world while Wn in the 43 poorest
countries have only about 5% grtr longevity.
A closer examination of life expectancy rates reveals that
there are also considerable variations in the differentials of
men & women among countries of similar SES levels.
Percent Difference in Life Expectancies at Birth of
Men and Women in Sub-Saharan Africa
Togo *l n aa
Chanr // ?? l' .S ___ _
Center. Air. Rep.
Niger G-e-- e
Mali ..W._. .
Rwandal " i. .
Chad u *>> >. .
Guinea-Biuau n e.....
Tanaumnis i. --t-a s-fo-hmn womn]heif
0 2 4 6 8 10 12 14
Percent Greater Longevity of Women
In sub-saharan Africa, which this chart shows, the mean difference
between the life expectancies of men and women is about 5%. In
Guinea, however, men and women have almost the same life
expectancy at birth while in Sierra Leone [which has the lowest life
expectancies in total years for both men and women] the life
expectancy of women is 15% greater than that of men.
These variations in gender differentials among countries in the
same geographical region and with roughly similar economic
conditions suggests that there are corresponding variations in the
relative status and living conditions of men and women between
these countries. The gender analyst would be particularly
interested in those countries that deviate significantly from the
mean; for example, they would want to know why women's life
expectancy is so low relative to men's in Guinea...and also why
men's life expectancy is low relative to women in Sierra Leone.
As previously noted, there are four countries in the world where the life
expectancy of women is only equal to or less than that of men according
the the 1992 World Development Report. These four countries are all in
South Asia, and as you can see in this map, they are all neighbors. They
are India, Pakistan, Nepal and Bangladesh.
These observations suggest that there is something about the cultures
of these countries that produces a devaluation of women that results in
lowered life chances for women relative to men.
Where men outlive women...
Before diving into an indepth cultural analysis of gender relations in these
countries, however, we can use quantitative data sources to help us refine
and focus our research questions. For example, in the countries of South
Asia, it would be informative and efficient to find out:
(1) At what ages are women's mortality risks greater than normal?
(2) do gender differentials in life expectancy vary by
a. rural and urban residence?
d. educational attainment?
* at what ages are women's mortality
risks greater than normal?
Sdo life expectancy differentials vary by
rural and urban residence?
This chart, for example shows the rates of maternal morality in the
four countries compared to the average of all low income countries
which is indicated by the bold horizontal line. We can see that
three of the four countries have maternal mortality rates well
below the group average. There is no discernable pattern seen here
to indicate that childbirth practices have anything directly to do
with gender imbalances in life expectancy in these countries.
Maternal Mortality Rates
(per 100,000 Live Births) of
India, Pakistan, Nepal &
0 nn,.an .
Source: World Development Report, 1993
ow I eAverag/- 576
Child Mortality Rates (per 1,000
Live Births) of India, Pakistan,
S Bangladesh, and Nepal compared to
Other Low Income Countries.
D Females 100-
I Males 0-80
Source: World Development 2 U .0
Report, 1993 m
In response to such questions, we see in this chart for example,
the sex-specific rates of childhood mortality for the four countries
and the average rates for all low income countries combined(left).
Childhood mortality is the number of children who die by the age of
five for every 1,000 population. Notice that, male child mortality is
greater than female mortality among the rest of the low income
countries [and almost everywhere else in the world], but in each of
these southern asian countries, female child mortality is higher
than the male rate.
This would lead us to suspect that female children are not as
highly valued and cared for as male children. And for insights into
this question we can look at data from the Demographic and
Health Surveys concerning the sex preferences of parents.
Sex Preferences of Women without Sons for Their
Next Child, Pakistan 1990-91.
Ia Male a Female No Pref.
0 = No children
1 = 1 child, no son
0 1 2 3 4 5
2 = 2 childrcn,no sons 4 = 4 children, no sons
3= 3 children, no sons 5 5 children, 0-1
This chart is adapted from data from the Pakistan DHS 1990-91.
It shows the sex preferences of women respondents for their next
child by the number of children they already have. In this chart,
only women who do not have a living son are shown. Clearly, among
women without sons, the desire for a son increases with the
number of children they have. Even more striking is the fact that, if
they do not report wanting their next child to be a boy, then they
are most likely to say that the sex of the next child makes no
difference. At none of the levels of this chart is the preference for
a girl large enough to even show up!
Even among women who already have one son, the preference for
boys is still highly apparent and that preference increases with the
number of girls they already have. Only among women who have
just one child and that child is a boy are there enough respondents
who want their next child to be a girl to show up on the chart.
Probably, a closer examination of DH5 data would reveal that
these women are among the more educated who plan on having only
two or three children so they want to have at least one boy and
In this section we began with the observation that in South Asia,
the life expectancy of women relative to men is shorter than in
other parts of the world. This led us to look at mortality rates of
different groups where we found that, unlike other countries, girls
under age 5 in South Asia are more likely to die than are boys. We
then linked these differential mortality rates to a strong prefer-
ence for sons in this region--Pakistan was the example used, but
son preference is also extremely high in the other countries of the
region. The question that arises from this preliminary investigation
that needs to be answered by a more indepth gender analysis is:
What are the institutional and cultural characteristics of these
societies that make sons so highly valued and daughters so lowly
valued to parents in these countries? An even more difficult, but
very important question is: once the culture that produces these
values is understood, how can development interventions be
designed and implemented to achieve better balance between
genders in this region?
We turn now to an examination of gender differentials in
educational attainment. We know from previous research that
there is a very strong positive relationship between educational
levels and life expectancy. We know, for example, that there is
significant improvements in family health associated with the
education of parents, particularly mothers. This association also
tends to hold regardless of families' socio-economic status.
An educated populace is also important for economic growth,
environmental protection, democratization, and for effective family
planning. Yet, a large portion of the developing world have little or
no formal schooling and in almost all developing countries, women
are less likely to receive schooling than men.
The most basic measures of education that are usually available
are sex and age disaggregated literacy rates. These rates tell us
what proportions of males and females in each age group have at
least minimal reading and writing abilities. These data would not
be sufficient for any indepth analysis of educational differentials in
developing countries, but they can give us a general idea of the
magnitude of the gender gap in education.
Here is a population pyramid of Nepal in 1991, one of the countries
in our previous example. Super-imposed on the total population, is
a graph of the literate population by age and sex (the light blue
sections). We can see that, not only is there still a relatively low
degree of literacy in Nepal, but also that men are much more likely
to be literate than women.
When we break the Nepali population down into rural and urban
components, we begin to get a better picture of where the
disparities lie. This chart also shows sex-disaggregated literacy
rates for 1981 and 1991 to give us some idea of what kind of
progress Nepal is making at improving literacy and gender equity.
All groups--rural and urban, men and women-- have increased since
1981. The gap between female and male literacy rates, however,
has not decreased for either rural or urban residents; and in fact,
the gap has widened noticeably in the rural sector. These obser-
vations suggest that women are still not receiving an equitable
share of benefits from development in Nepal.
In addition to educational data, there are also sex-disaggregated
data available on economic activity. We have to be careful in using
these data, however, because the traditional narrow definition of
economic activity has tended to under-estimate the contribution
of women to the economic system particularly regarding their roles
in agriculture and the informal sector of the economy. One of the
major goals of gender and development programs has been to
improve thequality of sex-disaggregated data so that problem
areas can be more accurately identified and developmental
progress can be better tracked.
I ICochabamba, Bolivia: Percent of
i............ Population Economically Active by
..... ....... ....... ..
70 soy Fepmalme
Rural a um
This is an example of data on economic activity in the department
the percentages of rural and urban men anve women classified as
mically ace in 6 and 11. Notice that e 19erentae
of active rural men has not change in 15 years an that the
Soupercentagepubli of Bactive urian men has actually ecline slightly.
ThiConversely, the percenof dtages of economically activitye women--whilent
ofstill not chaual to men olivia, from their 1992 crisenue. ramatically sincehow
the1976, particularly in the rural sector. Fromen and more detailed a
ecoanalysis of census fin 197dings and 1991.methodology, we concludtice that the percentage
of active rural men hao not changed in 15 yearg and that the
increased activity of women is due partly to better definitions of
the concept and partly to a real increase in female employment
relative to males.
Cochabamba, Bolivia: Primary and
Secondary Activities of Rural
a Care of Household aCommerce
a Construction, Transport & Services
Source: Cochabamba Rural
Household Survey, 1992
The simple dichotomy of either active or inactive, however, does not
fully capture the scope of gendered divisions of labor. These
charts are derived from a rural household survey conducted in
Cochabamba in 1992. They break down women's primary and
secondary activities in the two main climatic regions of this
department. The main point to be made here is that, while women
in the tropics and highlands are more likely to report care of
household their primary activity than any other type of labor, these
same women show much more variety in their secondary activities.
When we consider that most of the women who reported care of
household as their secondary activity, did some other kind of work
as their primary activity, we are drawn to the conclusion that well
over 80% of the rural women of Cochabamba are economically
active in some way and to some degree.
.. .. .. .. .
...........'' '' '
.... .... .... .... ...
. . . . . .
~s-.- ;s;sst;;;~r~;;;=~ ~
Our last examples are taken from Demographic and Health Survey
of Ghana, 1988 and they illustrate one way in which gender
considerations can be included in family planning programs. We
believe that understanding gender roles and relationships in
particular settings is an important pre-requisite to designing and
implementing successful family planning programs. Family
planning decisions are made within a marital context and involve
both husbands and wives, yet too often family planning strategies
have focused almost exclusively on the responses of wives to D & H
Traditionally, analysis of unmet need for family planning has
focused on the responses of women concerning their reproductive
aspirations and current contraceptive practices. This approach
assumes that women's aspirations and actions are shaped
independently of any marital context. This chart shows the
percentage of women in the survey who say that they want no more
children. In Ghana, we see that nearly a quarter of the women
desire to stop their fertility.
When we add the aggregate responses of husbands we see that
about twenty percent of the men also want no more children. From
these aggregate data, we might be led to assume that there is a
fairly high level of agreement among Ghanaian families and that
there is a demand for family planning services among at least one
fifth of the families.
When we match husbands and wives together in this survey,
however, a different picture emerges. We see that only 12% of
Ghanaian couples agree that they want no more children. 25% of
couples disagree on their desires for more children and 67% agree
that they want more children. The relatively high amount of
disagreement among couples in this survey suggests that
definitions of unmet need and solutions to addressing this familiar
problem need to focus more on the couple, and specifically on how
couples make decisions concerning fertility regulation.
Desire for No More Children, Ghana, 1988.
Source: Adapted from
Want no more children and Health Survey,
The 0 & H surveys do not provide much information on how family
planning decisions are made in households, but some questions
about communications between husbands and wives are asked
that can provide some helpful insights. For example, we see here a
chart showing that only 24% of the couples in the Ghanaian survey
agree that they have discussed family planning. 67% of couples
have never discussed family planning.
Fortunately, the need to more fully incorporate gender issues into
developmental research is beginning to be recognized. The DHS is
doing more husband surveys than it did before and is also testing
a set of questions in Nepal that are intended to gather informa-
tion on couple's decision-making processes.
We have shown several examples of how quantitative data can be
used as a catalyst for gender analysis. These data do not provide
answers to developmental problems. Rather, they identify
potential gender issues that warrant further examination and they
help to focus our course of inquiry by generating relevant questions
to be pursued through a more indepth gender analysis. They are
an important first step toward finding answers to problems of
sustainable development. The underlying premise of this
presentation has been that development practitioners need to
take into account the intricacies of the social reality of the
populations they are trying to help. They need to have a good
understanding of the relationships between genders, age groups,
ethnic groups and social classes of the social milieu within which
they are working in order to design efficient and effective inter-
vention strategies that will have people level impact. This means
that they need to develop a solid and extensive knowledge base in
which quantitative and qualitative data are integrated to provide
practical information for advancing the progress of sustainable