1. Workshop on Improving Gender Statistics in Rwanda Session 6 Analysis and Presentation of Gender Statistics Serena Lake Kivu Hotel, Rubavu District March 25-27, 2014

2. Learning Objectives
At the completion of this session, participants should understand and become familiar with:
how analysis and presentation of gender statistics can enhance the usefulness of the statistics;
the main types of analytic measures and analytic tools that can add value to basic data; and
tools and techniques for presenting statistics in ways that ensure the visibility of meaningful differences and similarities between women and men.
Primary references:
UNSD 2013, Integrating a Gender Perspective in Statistics, Chapter 4
UNFPA 2013, Guide on Gender Analysis of Census Data
UNSD and UNFPA presentations at the April 2013 UNSD Workshop in Japan
UNSD: United Nations Statistical Division
UNFPA: United Nations Fund for Population Activities
This session covers two distinct topics:
Gender analysis of statistics
Presentation of gender statistics.

3. Analysis of gender statistics
Analysis is an integral part of the statistical production process. In broad terms, analysis of gender statistics involves:
Identifying the gender issues to be informed by the analysis.
Obtaining statistics and other relevant data from available sources.
all variables of interest need to be disaggregated by sex as a primary classification;
many variables may also need to be cross-tabulated, e.g. labour force participation by sex by age group by geographic area.
Analysing and interpreting the data, including derivation of indicators and other analytic measures.
Reporting the findings, including presenting the statistics in easy-to-use formats that are appropriate to the statistical product in which they will be disseminated.

4. Key steps in analysing gender statistics
Identify gender issues
Obtain relevant data from available sources
Analyse and interpret the data
Report the findings

5. I. Gender analysis
The purpose of the analysis determines the type and level used.
The type and level of analysis usually vary by the type of statistical product to be used in reporting results.
Disseminating basic data collected in censuses and surveys typically involves tables with minimum data processing and analysis.
More analytical reports or articles typically conduct additional processing and analysis.
For most types of analysis, indicators and other analytic measures play an important role.
Use the basic data to select and construct relevant indicators and other analytic measures.
Apply more complex analytic tools and techniques to the basic data to better understand some issues.
Explain what is an indicator and how it differs from a statistic

6. Analytic measures
In analysing data from a gender perspective, use measures of composition and distribution of particular variables by sex.
Such measures include:
proportions and percentages;
ratios and rates;
medians and quantiles, means and standard deviations.
They provide the basis for constructing many of the gender indicators used to monitor progress towards gender equality.
Define all terms
Give examples of proportions or percentages, and what is the difference between them; ratios and rates, medians, quantiles, means and standard deviations
What is an indicator? When is it a gender indicator?

7. Proportions and percentages
In gender statistics, proportions and percentages can be calculated as relative measures of:
Distributions of each sex across the categories of a characteristic--e.g.,
Proportion or percentage of women who are employed--compared to women that are not employed or unemployed;
Labour force participation rate of women of the total female population;
Literacy rate of women—literate versus illiterate women.
Sex distributions within the categories of a characteristic-e.g.,
Proportion or percentage of the employed who are women or men;
Proportion or percentage of parliament members who are women or men;
Share of women (or men) among older persons living alone.
The following 2 slides illustrate these two types of measures
These two types of measures are illustrated in the table on economic activity status shown in the following two slides.

8. (a) Distribution of each sex across the categories of a characteristic
Distribution of employed males and females by occupational group
Rwanda. Occupational group by sex and urban/rural (percentages)
EICV3
Male %
Female %
Urban
Rural
Total
Professionals
3.5%
2.0%
7.1%
1.9%
2.7%
Senior officials and managers
0.1%
0.5%
0.0%
Office clerks
1.0%
0.8%
3.7%
0.4%
0.9%
Commercial and sales
7.7%
7.4%
15.7%
6.1%
7.5%
Skilled service sector
7.0%
4.0%
18.3%
3.2%
5.4%
Agricultural and fishery workers
61.3%
81.9%
34.5%
79.2%
72.6%
Semi-skilled operatives
12.9%
2.8%
12.3%
6.5%
Drivers and machine operators
5.2%
0.3%
6.4%
2.5%
Unskilled labourers
0.6%
Missing information
1.2%
100.0%
Persons aged 16 years and above who were working in the previous 12 months.
Sex is the characteristic and occupational group the categories
Distribution of unemployed males and females by occupational group:
Female and male totals are used as denominators, proportions calculated by columns
Used for comparison of employed women and men with regard to their occupational category
The basis for calculating gender gap: for example, the 2.8% proportion of employed women among semi-skilled operators is 10.1 percentage points lower than the proportion of men employed in this category, for a gender gap of 7.3%. On the other hand, women are much more likely than men to be working in agriculture as shown by the more than 20 percentage points difference between women and men employed as agricultural and fishery workers.
Source: Rwanda NISR, 2012, The Third Integrated Household Living Conditions Survey (Eicv3) Main Indicators report

9. (b) Sex distribution within the categories of a characteristic
Female and male distribution within marital status and education categories
Rwanda: Percent distribution of women and men age by selected background characteristics, 2010
Women
Men
Background Characteristics
Weighted %
Weighted number
Total Number %
% Men
Total %
Marital Status
Never Married
38.7
5,285
50.5
2,873
8,158 65 35
100%
Married
35.1
4,799
34.1
1,938
6,737 71 29
Living together
15.3
2,098
13.4
761
2,859 73 27
Divorce/separated
5.5
746
1.6 92
838 89 11
Widowed
5.4
743
0.4 22
765 97 3
Total
100
13,671
5,686
19,357
Education
No education
15.5
2,119
10.3
583
2,702 78
Primary
68.3
9,337
68.8
3,916
13,253 70 30
Secondary
14.7
2,008
18.7
1,064
3,072
More than secondary
1.5
207
2.2
125
332 62 38
5688
19,359
Note: Education categories refer to highest level of education attended, whether or not that level was completed
na = Not applicable
Female and male distribution within the categories of marital status and educational attainment:
Categories of the characteristic—such as never married or more than secondary--are used as denominators, and number of women and men as numerators; the proportions are calculated by row.
Used to show the under- or over-representation of women or men in selected marital status or education levels: for example, females account for 65.0% of never married persons and 78% of persons that have never attended school.
Source: Rwanda NISR, 2012, Rwanda Demographic and Health Survey 2010, Final Report

10. Rates
Rates of incidence can be used to study the dynamics of change over time.
They are a special type of ratio obtained by dividing number of events during a period by number of population exposed to the events during the period—for example:
Fertility rates – an average based on number of births -- in Rwanda, the total fertility rate (in the last 3 years) was 4.6 (2010 DHS)
Mortality rates, such as
Infant mortality rates (under 1 year) per 1000 live births – in Rwanda, 50 (2010 DHS)
Females: 55
Males: (2010 DHS),
Under five mortality rate per 1000 births – in Rwanda, 76 (2010, DHS)
Females: 97
Males: (2010 DHS)
By convention, some percentage measures are also called rates–for example:
Literacy rate (percentage of population that is literate) – in Rwanda,
Literacy rate among population 15 and older in 2010 was 70% (EICV3):
Among women 15 and older: 65% (EICV3)
Among men 15 and order: % (EICV3)
Fertility rates: expected number of children a woman o reproductive age (15-49 years old) would have during her life time if she experiences the given age at specific rate
Mortality rates: number deaths per a number of the population
Infant mortality rates: deaths before age 1 per 1000 live births
Under five mortality rates: deaths before age 5 per 1000 live births
Child mortality rates: under five mortality rate minus infant mortality rate – deaths per 1000 between 1 and 5 years
Maternal mortality ratios: maternal deaths per 100,000 live births

11. Ratios
Particular compositional aspects of a population can be made explicit by the use of ratios, where a single number expresses the relative size of two numbers– for example:
Sex ratio (number of males per 100 females): 93 for Rwanda (2012 Census)
Sex ratio at birth (number of male live births per 100 female live births): for Rwanda (2010 DHS);
Maternal mortality ratio: 496 for Rwanda (2010 DHS)
For some sex ratios, standardisation of the variables used may be necessary to adequately reflect gender differences– for example:
The gender parity index for primary gross enrolment, calculated as the ratio of the gross enrolment rate for girls to the enrolment rate for boys,
This rate controls for the sex composition of the school age population.
For Rwanda, the gender parity index for gross primary enrolment: (2011, Rwanda Education Statistics)
For some gender indicators based on sex ratios, standardisation of the variables used in calculating the ratio may be needed to adequately reflect gender differences.
For example, calculating the gender parity index for participation at various levels of education as the ratio of girls enrolled to boys enrolled gives a poor measure of gender differences in access to education because differences in the school age population of girls and boys are not taken into account.
An alternative calculation that controls for the sex composition of the school age population uses the ratio of the education enrolment rate for girls to that for boys.
Girls’ gross primary enrolment rate = girls enrolled as a proportion of school-age girls  129
Boys’ gross primary enrolment rate = boys enrolled as a proportion of school-age boys 126

12. Measures of central tendency and dispersion
Medians and quantiles
Often used in gender statistics to show the distribution of income or wealth across the population;
They can be useful in studying gender issues associated with poverty or in analysing the economic resources of different household types (such as single mother households).
Means (averages) -- Examples of gender-relevant indicators:
Average (mean) time use on unpaid work
Average (mean) size of land owned
Mean age at first marriage
Mean age of mother at first child
Standard deviations, coefficient of variation, etc.
These measures are important for measuring the degree of association between variables and making population inferences based on sample data.
Although not often presented in gender statistics
Projections
An example relevant to gender statistics is the projection of the male and female populations to a specified date in the future.
The size of the standard variation relative to that of the mean is called the coefficient of variation.
The ratio of women’s to men’s average earnings is a common indicator of the gender pay gap.

13. Understanding gender differences using analytic measures
It often may be necessary to disaggregate simple summary measures or combine them with other data to adequately inform gender issues. This is illustrated in the following example which explores poverty among female or male headed households in Rwanda.
Data from the 2012 EICV3 revealed little difference in poverty incidence among female-headed and male-headed households:
47% of female-headed households (which comprise 28% of all households), were poor compared to
44% of male-headed households (which comprise 66% of all households).
However, the data revealed a higher poverty rate for de facto female-headed households (51% compared to 47% for de jure female-headed and 44% for male-headed households). In these households, which are only 6% of all households, male heads were absent for more than 3 months in the previous 12 months mainly because of detention or compulsory service (41%) or for work (28%).
For households in extreme poverty, the difference by household head is much larger: 34% of de facto female-headed households are extremely poor, compared to 26% of de jure female-headed and only 23% of male-headed households. 
De facto female-headed households heads tend to be more similar to male than to other female heads in terms of marriage, age, education and literacy, but they are larger (6 members on average compared to 5 for male-headed and 4 for de jure female-headed households) and more likely to have younger children or grandchildren at home, and more likely to work as wage workers (32%) than de jure female-household heads (17%).
De jure female-headed households tend to be headed by older, widowed, less educated and literate women (almost 70% are 45 or older), who are much more likely to be small-scale farmers. These characteristics explains in part why they are different from male and de facto female-headed households.
Source: NISR. EICV3 Thematic Report on Gender, 2013.

14. Use and value of standardisation
In some situations it can be useful to standardise a measure to better understand gender differences or to avoid it being misleading (or biased).
Examples where standardisation may be important for the analysis:
Risk of renewed divorce of men or women in second or third marriages.
Standardisation by order of marriage can take account of the fact that more men than women remarry after a first divorce or widowhood.
Literacy rates of women and men.
Age standardisation can take account of the fact that literacy rates are lower at higher ages, were women predominate.
Incidence of disability in women and men.
Age standardisation can take account of the fact that there are more women than men in the population and that the excess of women over men is concentrated in the oldest ages where disabilities are most common.

15. A country example showing effect of age standardisation
Un-standardised and Age Standardised Prevalence of Selected Types of Disabilities in Mexico, based on 2010 Population Census
Notes on table:
Column 2
Percentage of persons with a disability who are women 50.1%.
Columns 3-4
Percentage of women who have a disability is 4.00%, i.e. the prevalence of disability among women. This compares with 4.17% for men.
Columns 5-6
These show the hypothetical results from age standardisation.
Age standardisation involves applying the percentage of males, and of females, with a disability by age to the same age distribution. In other words, rather than applying each percentage to the corresponding male or female population, each percentage is applied to a common population which, in this case, consists of all individuals of both sexes.
The age-standardised results vary between sexes because of the different proportions of disabilities among women and men, but not because of the different numbers of women and men in the base population.
The age standardised percentage of women with a disability is lower (3.87%), while the corresponding percentage for men is higher (4.29%). For further details, see the UNFPA Guide on Gender Analysis of Census Data, Chapter 12 Disability.
Source: UNFPA Guide on Gender Analysis of Census Data

16. Use and value of multivariate analysis
Multivariate analysis can
assist in disentangling variability and understanding interrelationships within a population group
provide a more comprehensive view of different relationships, making it easier to identify situations where, for example, the relationship between two variables can be accounted for by their common dependence on a third factor.
Examples of its use in the context of gender statistics are:
Understanding the relationship between women’s educational attainment and their economic level in rural and urban areas and at varying ages;
Investigating whether the relationship between two characteristics that are highly correlated, such as lower education and early marriage, is caused by another factor, such as belonging to a certain ethnic group;
Understanding whether the marital status of a woman has a direct effect on her labour force participation after controlling for other intervening factors.
Understanding the various factors that affect age of marriage;
Two types of multivariate analyses which have proved useful in social studies are
multiple linear regression and
logistic regression.
Multiple classification analysis (MCA) is another useful technique, closely related to linear regression.

17. A country example: A study using multivariate analysis in Rwanda
Measuring the Success of Family Planning Initiatives in Rwanda:
A Multivariate Decomposition Analysis.
Contraceptive use in Rwanda has increased far more than the Ministry of Health projected for the year Moreover, other indicators of progress, such as women delivering in health facilities and reductions in infant and maternal mortality have also exceeded expectations.
This study described the family planning initiatives in Rwanda and analyzed the 2005 and 2010 RDHS data to identify factors that contribute to the increase in contraceptive use, by decomposing the contributions of women’s characteristics and their effects.
The study found a mean predicted increase of in contraceptive prevalence rate between 2005 and 2010.
The largest increase (77 percent) results from changes in the effects of women’s characteristics compared with changes in these characteristics (17 percent).
The variables showing significant contribution in effects are women’s education, experience of child mortality, and place of residence.
As for compositional differences, woman’s education, exposure to family planning messages in the media or at health facilities, husband’s desire for children compared with wife’s, and woman’s child mortality experience had relatively greater effects.
Additional research is needed to assess the contribution of supply side factors that would have been also important for the increased contraceptive use in Rwanda.
Source: Muhoza, Dieudonné Ndaruhuye, Pierre Claver Rutayisire, and Aline Umubyeyi Measuring the Success of Family Planning Initiatives in Rwanda: A Multivariate Decomposition Analysis. DHS Working Papers, 2013, No. 94. Washington, DC: USAID.
Muhoza, Dieudonné Ndaruhuye, Pierre Claver Rutayisire, and Aline Umubyeyi Measuring the Success of Family Planning Initiatives in Rwanda: A Multivariate Decomposition Analysis. DHS Working Papers, 2013, No. 94. Washington, DC: USAID
The source publication provides details of the multivariate analysis of the factor that had an effect on the increase in contraception use.

18. Some tips for analysing gender statistics
Assess data quality to avoid misinterpretation of results.
Use appropriate analytic measures and techniques to construct indicators that reflect the gender issues to be studied.
Consider the value of using multivariate analysis to assist in understanding gender inequality in its many dimensions.
Interpret the results of analysis with careful consideration of the different factors that may be involved (such as distinguishing the between the impact of socio-economic and biological factors on health outcomes).
Take care when combining data from different sources and use appropriate techniques.

19. Concerns when integrating data from different sources
When different sources need to be combined to calculate a particular analytic measure (e.g., a rate), check the sources for consistency and comparability.
Comparability issues can arise because of:
differences in concepts, definitions, coverage or time period;
errors or variations in classification or data processing procedures; or
variations in concepts or practices in different years within the same source.
In most cases comparability checks can be made by reviewing each source’s documentation.
Consider consulting also the specialists who supply or use the data from that source.

20. Some further considerations ...
Be aware of the different implications, for gender analysis, of data produced at different levels of statistical unit.
Statistics on poverty may be produced at household level and/or individual person level but the concepts used are not the same and thus not comparable
Using sex of ‘head of household’ (or household headship) to analyse gender differences can be problematic.
‘Head of household’ can refer to many different concepts; it does not capture intra-household gender inequalities; and it can reinforce gender stereotypes.
There is no uniformity in country practices concerning the concept or its use.
Comparing households with different characteristics can provide useful insights into gender issues.
It can be useful to disaggregate households by
size and composition (sex and age of each member),
type (one person, couples with/without children, single mother/father with or without children, etc.) and
other characteristics
Poverty statistics at household level and individual level are discussed in UNSD Integrating a Gender Perspective in Statistics (Poverty section of Chapter 2).
Conceptual and practical issues associated with ‘head of household”, and types of household disaggregation that can provide insights into gender issues, are discussed in detail in UNFPA Guide on Gender Analysis of Census Data (Chapter 7: Households and Families).

21. Exercise 6.1: Data analysis
You are asked to prepare an article analysing a topical gender issue—on health, education, employment or violence--based on results of the 2012 Population Census.
How would you go about the task?
What analytic measures and tools would you expect to use?
Does the analysis of gender statistics need to be improved in any of the fields (topic areas) with which you are familiar? What are the specific field(s)? What needs to be done and why?

22. II. Presentation of Gender Statistics

23. Presentation of gender statistics
How the statistics are presented will influence understanding and use of the data for program or policy making
Gender data presentation seeks to:
Highlight key gender issues
Facilitate comparisons between women and men
Convey the main messages resulting from data analysis
Reach a wide audience
Encourage further analysis
Stimulate demand for more information
Tables, graphs and charts are the key forms of statistics’ presentation.

24. These are powerful ways to present data. They can:
Graphs and Charts
These are powerful ways to present data. They can:
Summarize trends, patterns and relationships between variables;
Illustrate and amplify the main messages of a publication, and inspire the reader to continue reading;
Give a quick and easy understanding of the differences between women and men.
A graph or chart should:
Be simple, not too cluttered
Show data without changing the data’s message
Clearly show any trend or differences in the data
Be accurate in a visual sense—for example
If one value is double another, it should appear to be double in the graph or chart.

25. Types of Graphs and Charts
There are many types of graphs and charts. It is important to select the right type for data being analysed.
The selection may also be influenced by the message to be conveyed and the method of dissemination (e.g. printed or electronic).
Some of the main types of graphs and charts used in presenting gender statistics are:
Line charts
Bar charts: vertical, stacked and horizontal
Age pyramids
Dot charts
Pie charts
Scatter plots
Maps

26. Line charts
Line charts can give a clear picture of trends over time—examples:
Trends in sex ratios;
Literacy rates over time;
Labour force participation rates by age group over time.
It is generally recommended that line charts start from zero at the y-axis of a variable. However, in this case, starting from an age of 15 and end at 49 makes more sense because very few births happen before the age of 15 or after the age of 49.
It is important to use a consistent scale on each axis, i.e, the distance between each number should be the same; otherwise the line’s shape can give incorrect impressions about the information.
Source: NISR & Measure DHS Rwanda Demographic and Health Survey (RDHS) 2010

27. Line charts (continued)
Rwanda: Age-specific employment rate in the last 12 months, by sex, 2010 (RDHS)
Line charts also can give a clear picture of differences across age groups.
This chart shows that in Rwanda in 2010:
Labor force participation rates were lower for women than for men at all ages.
The largest gender gap is at years of age.
Source: NISR & Measure DHS: Rwanda Demographic and Health Survey, 2010

28. Vertical bar charts
Figure Sex Ratios: Number of males per 100 females by age group (EICV3)
Both vertical and horizontal bar charts are common for presenting gender statistics.
A key feature is that the greater the value, the greater the length of the bar.
Examples of use:
total fertility rate by region;
antenatal care by urban/rural area;
proportion of women having third and higher order birth by education level.
Design note:
3-D visual effect will not change the main story, but it will make the graph unnecessarily complicated and potentially misleading.
Source: NISR EICV3 Main Indicators Report

29. Vertical bar charts (continued)
Grouped (or clustered) bar charts can present a particular characteristic for women and men at the same time, facilitating comparisons between them.
Rwanda: Lower secondary education Gross and Net Enrolment Rate from 2008 to 2012
Source: Ministry of Education, MINEDUC Education Statistics Yearbook

30. Stacked bar charts
Stacked bar charts illustrate data sets containing two or more categories
Most effective for categories that add up to 100 per cent.
Common problems:
Bars with more than three segments are difficult to compare from one bar to another;
one or more categories may be too short to be visible on the scale.
Design note: Category/categories of most interest should generally be placed at the bottom of the bars to facilitate the comparison.
In this case, the age group categories are ordered from lower to higher age with the lower age closer to the Y axis.
Source: NISR EICV Thematic Report Gender

31. Stacked bar charts (continued)
Rwanda: Poverty levels, by sex of household head, EICV3
Stacked bar charts are also used to present the distribution of a variable within the female and male population
Examples
the distribution of female and male deaths by cause of death;
the distribution of female and male school attendance.
The chart indicates that most close to half of households were not poor and a majority are either not poor or only poor.
However, more female headed than male headed households are likely to be extremely poor: .
This is particularly true among de facto female-headed households: more than 1/3 (34%) where extremely poor.
Source: NISR EICV Thematic Report Gender

32. Rwanda: Household composition (% household members) by sex of head
Horizontal bar charts
Rwanda: Household composition (% household members) by sex of head
Horizontal bar charts are often preferred when
many categories need to be presented (e.g. regions of a country), or
where categories have long labels.
May be preferred for showing some type of time use data, because the left-to-right motion on the x-axis generally implies the passage of time
The chart illustrates the different composition of households with different heads. For example:
de jure female headed households are likely to have children and grandchildren as well as other relative but do not have spouses. The head indicted is the female head herself.
De facto female headed households are most likely to have children but not grandchildren, although they do list a spouse which could be absent and may be listed twice as head and as spouse
No surprising given that they are a are majority of households, Male headed households’ characteristics are closest to those of all households. Note that they tend no to have grandchildren or other relations present.
Source: EICV3.
Note: composition estimates for de facto female-headed households include the absent head.
Other relation or in-law includes parent, sister/brother, adopted child;
Other includes servant and unrelated

33. Age pyramids
Age pyramids are useful for describing the age structure of a population and its changes over time.
Pyramids can use percentages instead of absolute numbers to highlight the age groups where women or men are over-represented.
The narrowing of the pyramid for both females and males starting at around age 30 is evidence that Rwanda’s population is still quite young.
The difference between males and females starting around age 40 reveals the higher mortality among men during the violence.
The large difference between the population up to age 64 and aver that age also reveals the effect of the genocide.
Source: NISR EICV3 Thematic Report Gender

34. Dot charts
Primary school net attendance rate for girls and boys by wealth quintile and urban/rural areas Yemen, 2006
Preferred to bar charts when presenting many categories or data points, because bars can become too thin and difficult to interpret.
Can convey a lot of information in a simple way without clutter.
May be vertical or horizontal.
Design notes:
This presentation highlights even more the gender gap
The gender-blind total has been removed from the graph to keep the attention on the gender gap
Source: UNSD presentation “From raw data to easily understood gender statistics,” at Workshop on Integrating a Gender Perspective into National Statistics, Kampala, Uganda December Data from Yemen Ministry of Health and Population, and UNICEF, 2008.

35. Pie charts
Used for simple comparisons of a small number of categories that make up a total.
Can illustrate the percentage distributions of qualitative variables and as an alternative to bar charts.
Using more than five categories will generally make a pie chart difficult to read.
Rwanda: % of type of earnings for currently married women and men, 15-49, 2010
Design note: A common error is to show too many categories in pie charts, resulting in labels that are hard to read or shares that are too narrow.
Describe key differences between women and men’s types of earnings
Source: NISR & Measure DHS Rwanda Demographic and Health Survey 2010

36. Scatter plots
School attendance rates for 6-17 years old by sex and state, India,
Often used to show the relationship between two variables.
Useful when many data points need to be displayed, such as a large number of regions, sub-regions or countries.
Also useful for identifying outliers in the data.
Design note: the four states where girls have significantly lower school attendance rates than boys have been highlighted.
The dots that are close to the diagonal represent the states where the proportion of girls attending school is similar to the proportion of boys attending school
Source: UNSD presentation “From raw data to easily understood gender statistics,” at Workshop on Integrating a Gender Perspective into National Statistics, Kampala, Uganda December Data from India Ministry of Health and Family Welfare, Government of India, 2007

37. Maps
Maps are used to show spatial patterns and geographic distributions for a particular variable.
They can increase the visibility of regional clusters within a country and highlight regional pockets that deviate substantially from the norm.
The chart shows what proportion of the households in each province has access to safe drinking water. While the data are not disaggregated by sex, they are gender –relevant. Why?
Source: NISR EICV3 Main Indicators Report

38. Interactive graphs and charts
A range of data visualization tools can be employed to enhance on-line dissemination of graphs and charts.
These tools can animate presentations, provide other interactive features, and display three or four dimensions of data simultaneously. For example:
A moving image can show transitions in a variable over time—changing the shape of an age pyramid;
Actual values and other details underlying a particular point in a graph or chart can be displayed instantly on request--by hovering over the point;
Bubble charts (a variation of the scatter plot) can be used
to visualize three or four dimensions of data and also
be animated to show changes over time.
Bubble charts visualise the first two dimensions of data using coordinates on the x and y axes and represent the relationship by bubbles.
The size and colour of the bubble can then be used to represent two further dimensions of data.
Such a chart can display, for example, a country’s fertility rate and life expectancy at birth (through the x/y co-ordinates), as well as its population size (through the size of the bubble) and its region in the world (through the colour of the bubble).

39. Tables Tables are a necessary form of presentation of data.
Types of tables:
Large comprehensive tables, often placed in a separate part of a publication (e.g., in an annex).
Text tables, which are smaller and part of the main text of a publication. They often support a point made in the text.
Always preferable to presenting many numbers in the text itself, as they allow more concise explanations.
Selection of data to present should focus on most striking differences or similarities between women and men.
Some data may be more easily conveyed in a table than in a graph--For example,
when data do not vary much across categories of a characteristic, or
when data vary too much.
Many statistical publications have as a main objective the dissemination of data collected and have to be specific about the values observed for the characteristics measured. This can be achieved in large, comprehensive tables. For example, a single table may contain information on several characteristics and indicators and provide several breakdowns of variables.

40. Text tables with one column Rwanda: Fertility by province, 2010 DHS
Can be used to present data with not much variation between categories,
Often listed in ascending or descending order.
Province
Total fertility rate* (average # of children per woman)
City of Kigali
3.5 
North
4.1
South
4.6
East
4.9
West
5.0
Total (average)
*Total fertility rate for the 3 years (1-36 months) preceding the survey
Source: NISR & Measure DHS RDHS 2010.

41. Text tables with two or more columns
Used to present data for females and males side by side so differences are clearly visible.
Source: NISR EICV3 Thematic Report Gender.

42. Text tables with two or more columns (continued)
Can be used as a form of presentation when the focus of analysis is a breakdown variable
in this case, mother’s age--associated with a number of related indicators expressed in different units
Note how well the information about the variables in the table is provided. This is what metadata does and how it is used
Source: NISR & Measure DHS RDHS 2010

43. Some tips for user-friendly presentation of gender statistics
Focus on a limited number of messages for each table, graph or chart.
The messages should generally relate to a specific gender issue.
Adopt good design practices; for example:
Ensure charts have clear, simple headings; labels are clear and accurate; axes are clear and divided consistently; a key is provided; data sources are acknowledged.
Facilitate comparisons between women and men; for example:
Present data for women and men side by side;
Ensure consistency in the way data for women and men are presented (e.g., use the same color for women and men in all charts in a presentation).
Consider the audience; for example:
Rounded numbers may communicate a message more easily to the general public.
Ensure simplicity of the visual layout; for example:
Labels for values presented inside a graph or chart can be distracting and often may be redundant;
Including a third dimension on a two-dimensional graph/chart can be misleading.

44. Exercise 6.2
1. You are asked to prepare an article presenting the results for a topical gender issue based on the most recent Population Census—such as boys’ increasing dropout rates compared to girls’ permanency or women’s much lower access to and use of credit for agriculture
How would you go about the task?
What presentational tools would you expect to use?
2. Does the presentation of gender statistics need to be improved in any of the fields (topic areas) with which you are familiar? What are the specific field(s)? How would you improve the presentation and why?