Vietnam Module 2

by user








Vietnam Module 2
Gender Statistics Training Workshops: Vietnam
From Gender Issues to Gender Statistics
February 18-20, 2014: Moc Chau – Son La
February 25-27, 2014: Danang
Objectives of Session
The main objectives of this session are to:
highlight the key role of gender issues in gender statistics activity;
raise awareness of the range of data sources that can be used in producing
the statistics;
illustrate the importance of gender statistics by showing how they can
inform gender issues in several key areas; and
provide some insights into the conceptual and measurement factors that
can affect the usefulness of the statistics.
Primary references:
UNSD 2013, Integrating a Gender Perspective in Statistics, Chapter 2;
UNECE and WBI 2010, Developing Gender Statistics: A Practical Tool, Chapter 3.
How can gender issues be brought into statistics?
To adequately reflect gender issues, statistics need to:
• provide disaggregation by sex;
• focus on areas of concern where women and men may not have
equal opportunities or status, or their lives are affected differently;
• take into account specific population groups where gender inequality
is likely to be present or more pronounced;
• use concepts, definitions and measurement methods that can reflect
women’s and men’s status, gender roles and relations in society;
• be obtained from sources where the collection tools take into account
factors that might introduce gender bias into data.
How can the priorities for gender statistics be identified?
1. It is important to start by identifying and understanding the
gender issues and the data needed to address these issues
– describing the issues in terms of policy-relevant questions (e.g.,
‘do women earn less than men’) can help in determining what
data are needed.
2. Then assess existing sources in terms of data availability and
3. Based on this information, data gaps and their statistical
implications can be identified.
How can the priorities for gender statistics be identified?
4. These statistical implications may include:
- better utilisation of existing data, such as through recoding, retabulation or re-analysis of micro data;
- improvement in methodology of existing data collections;
- new data collection, such as a new collection instrument or
additions to an existing collection instrument;
- improvement in data dissemination.
5. Priorities in developing gender statistics should be set based
on this information and available human and economic
Country priorities in gender statistics
• National priorities for gender statistics depend on current policy goals
and plans as well as current statistical capacity. Different countries
have different priorities.
In Vietnam, for example, the Government has a National Strategy on Gender
Equality for 2011-2020. The strategy provides goals, objectives, and targets to
achieve equality between women and men in terms of opportunity, participation
and enjoyment in the fields of politics, economy, culture and society. To support
monitoring and evaluation of progress on the strategy, and to support other
needs for gender statistical information, a set of 105 National Statistical
Indicators on Gender Development has also been promulgated.
• Some gender statistics are produced by all countries. But there is
often a gap between statistics needed to address national gender
goals and statistics currently produced.
Country priorities in the ESCAP region
Findings relating to 20 countries (including Vietnam) in the ESCAP region that
participated in the 2012 UN Global Review of Gender Statistics:
Almost half the ESCAP countries reported that they had national priorities related to
gender statistics
Top priorities were:
- raising awareness of gender
- national plans/goals for gender
- gender-focused survey
- gender database development
- gender-related data dissemination
Priority thematic areas were:
- time use survey
- women in decision making
- women in entrepreneurship
- gender-based violence
- early marriage
What data sources are used in producing gender statistics?
• Many data sources are used by countries to produce gender
statistics. The main types are:
– Population and housing censuses
– Population sample surveys
– Business censuses and surveys
– Administrative records
• To construct some gender indicators, data from more than one
source may need to be combined.
• In general, no one source provides better or more genderrelevant data than others.
What data sources are used in producing gender statistics?
• The quality of gender-relevant data from each source depends on
many factors, including:
–concepts, definitions and classifications used
–collection design and coverage
–the way questions are asked
–the collection methods used
• Gender bias can be present in any type of data collection and arise at
any stage of the statistical production process.
• It is important to understand the value and limitations of the particular
sources available in a country for purposes of producing gender
Population and housing censuses
• Population censuses are typically the largest statistical collections undertaken
by a country and are conducted relatively infrequently (e.g., 10 yearly).
• They obtain data on each person in the population, including their sex, age
and other characteristics.
• Generally a range of other topics are also covered.
– For example: female and male labour force participation; current occupation; paid
and unpaid work; income; educational participation and attainment; aspects of
health and disability; living arrangements.
• They provide:
– a rich source of information for examining differences between females and males
across many dimensions of life and in fine geographical detail.
– information for studying families, households and population sub-groups from a
gender perspective.
– population benchmarks for constructing indicators and other analytic measures
for studying gender issues.
Population sample surveys
• Household surveys and other population surveys collect information directly
from individuals.
• They can cover a very wide range of topics in depth.
• Data collected invariably includes sex and age of each individual covered in the
• Surveys may have:
– a multipurpose focus, with many discrete topics
– a more general social focus, with a range of topics for analysis of cross-cutting issues
– a primary focus on a particular topic or population group (e.g., labour force, education, literacy,
health, disability, time use, domestic violence, migrants)
– attached modules on separate topics (e.g., modules on discrete topics may be attached to
regular labour force surveys)
• The implications of sampling error (as well as non-sampling error) need to be
considered when using data from these surveys to produce gender statistics.
Business censuses and surveys
• These collections obtain data from businesses and other organisations, as
well as from the registers on which the collections are based.
• The focus of the collections may be:
– particular industries or activities (eg manufacturing, agriculture, education services)
– economy-wide
• They can provide gender-relevant information if sex-disaggregated data are
collected for individuals associated with the organisation.
– For example: earnings of different categories of employees; characteristics of
owners or managers of businesses and agricultural holdings, including type, size
and location; numbers of students and staff in different fields at educational and
research institutions.
Administrative records
• These records contain data routinely collected through administrative processes.
• Where records hold information on individuals, including their sex, they can be a
valuable source of gender statistics.
• A wide range of topics may be covered.
– For example: school enrolments; registered unemployed; registered births, deaths and
marriages; registered diseases; use of health services; provision of income support;
arrests for criminal activity.
• They have the potential to provide more frequent, reliable and finely
disaggregated data than sample surveys
• But, their usefulness may be limited because their primary focus is administration
not statistics:
– For example: coverage, definitions, classifications and collection methods may be
deficient for purposes of gender statistics, and details held may not be current.
Selecting the right data source
• It is crucial to start with a good understanding of what data are needed and
for what purposes.
For example, data may be needed to produce a specific statistical indicator for
monitoring progress towards a gender equality goal. Relevant concepts, coverage,
disaggregation, frequency, level of accuracy etc. should all be clear.
• Existing and potential sources should be identified and evaluated in terms
of their ability to provide the required data at acceptable quality and cost.
For example, aspects to consider include: alignment of concepts, definitions and
classifications used by the source with those required; adequacy of population and
topic coverage; appropriateness of collection methodology; impact of sampling and
non-sampling error, including possible gender bias; flexibility of the source if
modifications are needed; and frequency and timeliness of data updates.
• If two or more data sources can provide data of acceptable quality and
cost, these sources should be further examined to determine the most cost
effective source.
Choosing between data sources: illustration of issues
Unemployment of women and men
• Regular labour force surveys may provided the most precise measures based
on international statistical standards, and data may be cross-classified by a
range of other variables.
• Other population surveys may provide less precise data, less frequently and
with less complete population coverage, but the data may be cross-classified
by additional variables of relevance for gender analysis.
• Population censuses may provide basic data for all the population with
disaggregation by fine geographic areas and small population groups, but
data updates may be many years apart.
• Administrative records may provide regular data on persons registered as
unemployed for purposes of receiving unemployment benefits, but
definitions and coverage may differ from those needed for statistical
purposes and data continuity may also be an issue (e.g.,, as eligibility rules or
administrative processes change over time).
Choosing between data sources: illustration of issues (continued)
Violence against women and men
• Personal violence surveys, when conducted as dedicated standalone
population surveys, generally provide the most reliable and detailed data on
prevalence of violence, its nature and consequences, and characteristics of
victims and perpetrators, but data may be infrequent.
• Other types of population surveys (e.g., crime victimisation surveys or
violence modules attached health surveys) may provide less detailed data but
more frequently and at lower cost. Coverage of victims and incidents may also
be less complete and non-response to sensitive questions may be higher,
possibly with some gender bias.
• Administrative records (e.g., police records, or records on use of health or
medical services) may provide regular data on violence reported to
authorities, but incidents of violence are often unreported (especially
domestic violence) and involve no services. Incidents reported to police may
also be unrecorded, e.g., if not formally investigated or not perceived as
crimes under criminal law.
Combining data from several sources
• Data from one source sometimes need to be combined with data from
another source to produce particular statistical indicators.
– For example, a prevalence rate may be calculated by dividing a population survey
estimate of a variable by a population estimate based on the last population census.
– It is important that the data being combined are compatible, eg. relate to the same
time period, have the same scope, and refer to the same statistical unit.
• Data from two different sources may sometimes be brought together to
form an enhanced data set.
– For example, data from a population survey may be combined with data from an
administrative collection or a population census by using data linking or matching
– An enhanced data set can add considerable analytic value, but it can be very complex
and challenging to create, particularly from the perspective of methodology.
Protocols may need to be established to cover aspects such as data ownership and
privacy protection.
Combining data from several sources (continued)
• Longitudinal data sets can be created by combining different waves of a
panel survey, selected records from different rounds of a population
census, or selected records from different time periods covered by an
ongoing administrative collection.
– For example, details reported by the same woman or man over a number of
years can be examined, allowing topics such as labour and income dynamics,
or family formation and labour force transitions, to be studied in depth from
a gender perspective.
– Longitudinal data can be very valuable in understanding individual
behaviours and pathways over time. However, as the population changes
over time, estimates for a given period may not represent the state of the
full population for that period.
Evaluating data sources for statistical indicators: a country example
In Vietnam, guidance has been prepared on issues to be considered in reviewing
potential data sources for producing the country’s gender statistical indicators
(ratified by the Vietnamese Government in 2011). These issues include:
– Orthodoxy of data. Only official data from government statistics agencies should be
– Consistency and comparability. Changes in concepts or collection methods over time
should be checked. When combining sources, comparability issues can arise from
differences in collection scope, duration, concepts and definitions, and collection
– Discussion with data producers. This can help to understand more clearly and interpret
correctly the meaning of data.
– Consultation with experts in different fields. This is essential to determine the right
data source, as they understand their data and the data’s status.
– Data releasing period. The frequency of a collection and time lag on releasing results
can affect the feasibility of using it for indicator purposes.
Bringing gender issues into specific fields of statistics
The next part of this presentation illustrates some of the gender
issues, data needs and measurement challenges in four specific fields:
- education;
- work;
- health; and
- population, households and families.
Each of these four fields is represented by a number of indicators in Vietnam’s
National Gender Indicator System. For each indicator, the system specifies its
frequency, various disaggregations, and the responsible agency.
Keep in mind the key steps in bringing gender issues into statistics –
they apply in these as well as all other fields.
Key steps in bringing gender issues into statistics
Identify gender issues
Identify data needed
Identify and assess sources of data
Identify and address conceptual and
measurement issues
Obtain data and produce statistics
EDUCATION: What are some of the gender issues?
A few examples relevant to Vietnam ....
• Do women have lower literacy rates than men? What are the trends? How are
gender differences in adult literacy distributed across geographic areas and
population groups?
• Is the male rate of secondary school enrolment much different to that of females?
How much do rates vary between urban and rural settings? What impact does
poverty have on female and male enrolments?
• To what extent does school attendance vary between females and males of
differing ages? What drives gender differences in dropouts?
• Is household education expenditure higher for boys than girls? To what extent
does spending vary by geographic location of the household?
• Are women under-represented among teaching staff? Have patterns changed over
EDUCATION: What can statistics tell us about these issues?
In the case of Vietnam, a 2012 report by the GSO included the following findings:
• The literacy rate of people 15 years and over is quite high, but the female rate is lower
than the male rate. This gap is higher in rural areas and ethnic minorities. The rural
female rate declined during 2006-2010 and is lowest among the poorest.
• Net enrolment rate of both lower and upper secondary school males is lower than the
female rate, particularly in rural areas and the poorest group of the population. The gap
has also tended to increase.
• Rate of not attending school for 6-14 year old females is higher than the rate for males
of this age. But the dropout rate of 14-17 year old males is higher than that of females,
showing the tendency of males to move earlier into the workforce.
• Average education expenditure of households per school person in urban areas is
double that of rural areas and increased rapidly over 2004-2010. Average expenditure
on males is higher than that on females in both urban and rural areas.
• The female share of teachers increased continuously during 2001-2010, reaching 41%
for professional secondary school and 47% for tertiary education.
EDUCATION: What types of data do countries generally need to inform
gender issues?
Data needed:
• Number of students in primary, secondary and
tertiary education, e.g., enrolments, completions etc
• School attendance and reasons for not attending
• Tertiary education graduates
• Education expenditure of households
• Number of teachers and researchers
• Number of schools with particular facilities
• Literacy
• Highest level of education attained
• Participation in non-formal education and training
• Participation in continuing vocational training
• Use of information services (e.g., for farmers)
• Users of the internet, computers, mobile phones etc.
• By sex, and often by
age as well
• Also important may
be level of education,
field of study, small
areas, and small
population groups
EDUCATION: Conceptual and measurement issues
Some common issues that may arise in producing gender statistics ...
• Educational participation. Numbers enrolled may overstate to different
degrees the educational participation of girls or boys, as persons enrolled
but not attending are included.
– Both enrolment and attendance statistics need to be considered.
• Population groups. Some groups with distinct gender differences in
educational participation may not be covered in statistics on enrolment or
attendance, resulting in biased estimates. Excluded groups may be those
outside the regular education system in the case of administrative records;
those living in remote areas or institutions in the case of household
surveys; or those studying abroad in the case of both sources.
– The magnitude of under-coverage should be assessed and users should
be informed of the implications for gender studies. Using a
combination of sources may improve coverage.
EDUCATION: Conceptual and measurement issues (continued)
• Literacy. Statistics based on self-reporting or proxy-reporting may
overestimate literacy rates, particularly for persons considered dependant.
– Direct assessment can provide more objective measures.
• Researchers. The proportion of researchers that are female or male may
differ between institutional sectors.
– All sectors should be covered for a complete picture.
WORK: What are some of the gender issues?
A few examples relevant to Vietnam ....
• Do females and males have different rates of labour force participation? Are
there gender differences in participation across different population groups?
Are women over-represented in the informal sector?
• Do male workers tend to have higher educational qualifications than female
workers? What are the trends?
• Are hours worked similar for males and females?
• Do females spend more time on housework than males? Does the pattern
differ between urban and rural areas?
• Are the earnings of women workers less than those of men? Is the gap
declining or growing over time? Which groups have the largest disparities?
WORK: What can statistics tell us about these issues?
In the case of Vietnam, a 2012 report by the GSO included the following findings:
• Overall labour force participation is high, and the female rate is much the same
as the male rate. For females, the highest rate is among ethnic minority women.
The majority of labour in the informal sector is female.
• Educational qualifications and technical expertise are higher, on average, for
male workers than female workers. While the proportion of females with
tertiary degrees grew during 2002-2010, the proportion having no qualifications
was almost unchanged. The proportion of untrained female workers is very high.
• Average working time per week for paid male and female workers is almost the
• Average time doing housework per person per day is much higher for females
than males, and the gap in rural areas is less than in urban areas.
• Average monthly income of paid female workers is lower than that of paid male
workers, and the gap grew in recent years. The largest income disparity occurs in
the untrained group and the 55-60 age group.
WORK: What types of data do countries generally need to inform
gender issues?
Data needed:
• Labour force participation
• Employment, unemployment, underemployment
• Status in employment and informal employment
• Number of hours worked
• Wages or earnings from work
• Ownership and management of agricultural
resources, use of agricultural inputs
• Time use by type of activity
• Access to and use of flexible working
• Availability and use of formal childcare services
• Maternity and paternity leave benefits
• Children in employment and in unpaid housework
• By sex, and often by
age, occupation, and
industry of activity.
• Also important may
be small areas, and
small population
WORK: Conceptual and measurement issues
Some common issues that may arise in producing gender statistics ...
• Labour force participation and employment. Women’s participation in labour
force and employment may be underreported. Reasons include: difficulty in
separating activities that should be included from those that should not;
gender-based stereotypes of women; the difficulty of capturing seasonal and
intermittent activity; and various coverage limitations when data are sourced
from business surveys.
– Collection methods and training may need to be improved. An alternative
data source may need to be considered.
• Unemployment. Women’s unemployment may be underreported. Reasons
include: women may be perceived or define themselves as not seeking work;
women are more likely to be discouraged or seasonal workers; and coverage
limitations when data are sourced from administrative records.
– Collection methods and training may need to be improved. An alternative
data source may need to be considered.
WORK: Conceptual and measurement issues (continued)
• Occupation and status in employment. These aspects are often not
recorded with enough detail to properly assess gender differences in
forms of work and employment conditions. Also women may be
misclassified in status in employment categories due to misclassification
of jobs.
– Collection methods and training may need to be improved.
• Gender pay gap. This may be higher or lower depending on the concept
used (e.g., wages of paid employees; earnings of paid employees including
overtime and regular bonuses; or income related to employment of all
workers, including all bonuses and social security benefits).
– Users should be informed of the particular concept used: various
concepts may be useful.
WORK: Conceptual and measurement issues (continued)
• Small agricultural holdings. The exclusion of small holdings from agricultural
censuses and surveys induces a gender bias in the statistics obtained as
women holders tend to concentrate in this sub-sector.
– The extent of the bias should be quantified and users informed.
• Productive agricultural resources. Comprehensive coverage of gender issues
in access to productive resources in agriculture requires use of data collection
units and data analysis more disaggregated than the holding level.
• Head of agricultural holdings. Many of the ‘male-headed’ agricultural holdings
may in fact be holdings headed jointly by women and men that are incorrectly
recorded due to omissions and gender bias of interviewers and/or
– Interviewer training and/or respondent instructions may need improvement.
WORK: Conceptual and measurement issues (continued)
• Work on own account production of services. Some forms of work are not
covered by conventional labour force statistics, as the statistics are limited to
activities which contribute to the production of goods and services as
defined by the SNA. In particular, own account production of services, which
is mostly carried out by women, is excluded.
– Time use statistics can shed light on both the included and excluded activities
performed by women and men. Examples of excluded activities are: preparing
meals, cleaning and other housework; caring for children and others in the
household; and directly-provided volunteer services.
– In particular, time use statistics can provide measures of unpaid housework
undertaken by women and men, provided that contextual information (e.g.,
whether the work was paid or unpaid and for whom it was performed) is
collected and that simultaneous activities are recorded.
HEALTH: What are some of the gender issues?
A few examples relevant to Vietnam ....
• What are the trends in infant mortality among boys and girls? Do the trends
differ between urban and rural areas? Is the gender gap narrowing?
• Is pregnancy-related maternal mortality decreasing?
• What is the prevalence of malnutrition among boys and girls? Is there an
association between child malnutrition and a mother’s level of education?
• Do females have as much knowledge of HIV prevention as males? What
groups have least knowledge?
• Is the prevalence of smoking among males declining? What gender
differences exist in tobacco use?
HEALTH: What can statistics tell us about these issues?
In the case of Vietnam, a 2012 report by the GSO included the following findings:
• The infant mortality rate rapidly decreased during 2001-2010, particularly in rural
areas. The rate for girls was lower than for boys.
• The pregnancy-related maternal mortality rate has sharply reduced in recent
• The malnutrition of children under 5 years old is quite high but there is no big
difference between boys and girls. There is an inverse relationship between the
prevalence of malnutrition and the educational attainment of the mother.
• The knowledge of HIV prevention is better among young males than young
females. Poor women are less exposed to knowledge about HIV than other groups.
• The prevalence of smoking is still high, with the rate among males very much
higher than the rate among females.
HEALTH: What types of data do countries generally need to inform
gender issues?
Data needed:
• Live births, place of delivery, delivery attendance
• Deaths and causes of death, abortions
• Children ever born and children surviving
• Weight and height of persons, vaccinations
received, selected health conditions and
treatments received
• Health expenditure of households
• Pre natal care, contraceptive use
• HIV/AIDS: prevalence, tests done, deaths,
knowledge, access to retroviral drugs, condom use
• Health risk factors, e.g., alcohol, tobacco, obesity,
physical activity, high risk sexual behaviour
• Unintentional and occupational injuries
• Population by sex and age for calculation of rates
• By sex, and generally
by age as well
• Also important may
be small areas, and
small population
HEALTH: Conceptual and measurement issues
Some common issues that may arise in producing gender statistics ...
• Gender gap. A distinction needs to be made between biological and social
factors to understand gender gaps. These factors may also be entangled.
Certain measures and indicators (e.g., of child mortality or nutrition) may
make biological factors less relevant than social factors, or vice versa.
• Infant and child mortality. Ascertaining sex differentials is difficult in
countries lacking a complete and accurate civil registration system for deaths.
Estimates of sex differentials based on household surveys may have large
standard errors and wide confidence intervals.
– Information on quality of the available data should be provided to users.
• Maternal mortality. Reliable data are lacking in many countries due to
underreporting and misclassifications of deaths. Also estimates obtained
from household surveys have wide confidence intervals.
– Where data on maternal mortality are suspected to be inadequate, it is
important to interpret the data within the context of other maternal
health indicators.
HEALTH: Conceptual and measurement issues (continued)
• Births and deaths. Data from censuses or surveys, as well as from civil
registrations, may have coverage and accuracy deficiencies. For example,
female births may be more severely underreported than male births in
countries where women have a lower status. Births and deaths may also be
underreported due to premature death or omissions as a result of proxy
responses or recall errors.
– Data quality should be assessed using multiple sources. User should be
informed about quality deficiencies.
– Adjustments may be needed for underreporting and for distortions in
the age structure.
– A data improvement strategy may be needed.
• Causes of death. Causes are often not reported, or misreported, for both
females and males in civil registration systems.
– Systematic and targeted efforts to improve reporting may be needed.
HEALTH: Conceptual and measurement issues (continued)
• Abortions. Reliable statistics are not readily available.
– Research may be needed to improve estimation methods.
• Nutrition. Sex differentials may be clearer when data on weight and height
of girls and boys under 5 are disaggregated by age.
• HIV prevalence. Sex bias may occur in estimates based on population
surveys due to gender differences in participation in testing.
• Sexual behaviour. Sex bias may occur due to normative reporting of sexual
behaviour, e.g., condom use and high risk sex . Use of contraceptive
methods may be underreported.
– Data sources may need to be reviewed, or existing methods adjusted.
• Alcohol consumption. Type and frequency of consumption may vary by
gender and surveys may not adequately distinguish the relevant risk
– Survey questionnaires may need to be improved.
What are some of the gender issues?
A few examples relevant to Vietnam ....
• Is the sex ratio at birth moving closer to, or further away from, the natural
level? Is there evidence of foetal sex selection?.
• To what degree does age at first marriage differ between women and men?
What impact does location (urban/rural) or ethnic background have on this
• Is family planning mainly undertaken by females or males? Have roles and
practices changed over the years?
• Are females over-represented among single persons? How does the female
share vary across age groups?
• What are the trends in household size and their gender implications?
What can statistics tell us about these issues?
In the case of Vietnam, a 2012 report by the GSO included the following findings:
• The sex ratio at birth (number of males per 100 females) has continuously increased beyond
the natural level, with the ratio in rural areas exceeding that in urban areas. This implies that
foetal sex selection (favouring males) is quite common and becoming more popular.
• Average age at first marriage in 2010 was higher for men (26.2) than women (22.7). In
urban areas both men and women marry later than in rural areas. Early marriage (15-19
years old) is not popular but has tended to increase in recent years and is quite common
among ethnic women.
• The task of family planning is mainly undertaken by women. Male participation is low and
seems unchanged in the last decade.
• The proportion of single persons that are female is higher than the proportion that are
male, particularly among the elderly.
• Average household size has declined in recent years, showing that the nuclear family is
quite popular.
POPULATION, HOUSEHOLDS and FAMILIES: What types of data do
countries generally need to inform gender issues?
Data needed:
• Live births, children ever born
• Population composition, including age and sex
• Marriages and divorces, duration of marriage
• Marital status, consensual unions
• Contraceptive use
• Females in reproductive age group
• Household type
• Young persons by household type
• Older persons by household type
• Older persons living in institutions
• Family nuclei of lone parents with young children
• By sex, and generally by
age as well
• Also important may be
geographic area,
urban/rural area, ethnicity,
migration status, wealth
status, educational
attainment, and other
variables relevant to
understanding living
Measurement Issues
Some common issues that may arise in producing gender statistics ...
• Household members. In some population censuses and surveys, female
members of the household may be more likely to be underreported than
male members.
– Collection methods may need to be improved and special promotional
material developed.
• Household type. Classifications of household type may need adjustment to
identify certain types of living arrangements that are most relevant from a
gender perspective.
• Informal unions. These may not be adequately covered in statistics, as the
marital status of an individual is usually recorded in relation to the marriage
laws or customs of the country.
– Special survey questions or reporting instructions may be needed to
capture data on informal unions in countries where they are common.
Measurement Issues (continued)
• Non-marital fertility. Data may not be available or detailed enough to
understand trends.
– Additional survey questions may be needed.
• Family planning. Unmet need for family planning has often not been
calculated using a comparable methodology over time. Use of contraceptive
methods may be under-reported.
– It may be necessary to standardise methodology and assess reporting
SUMMARY: What can be done about conceptual and measurement
issues in gender statistics?
• Understand the issue
impact on identifying gender gap or not? (every data source has limitations)
incomplete coverage? sex bias in questions? proxy respondents? non-response?
inadequately-trained interviewers? data coding, data entry or editing errors?
• Improve future data collection activities
• Adjust the estimates
imputation? benchmarking?
• Use multiple data sources?
confront the data?
combine the data?
• Use different methods
• Communicate with users
ensure users understand the quality of the data, including possible gender biases
provide reliability measures, e.g., statistical significance, confidence intervals
Fly UP