Annex B. Methodology
Variables included in the SIGI 2019 were selected according to the following criteria:
Conceptual relevance
The variable should be closely related to the conceptual framework of discriminatory social institutions and measure what it is intended to capture.
Data quality, reliability and coverage
The variable should be based on high quality, reliable data. When micro-data were available, the data have been harmonised and standardised across countries/territories. The data included should have extensive coverage across countries/territories.
Distinction
Each variable should measure a distinct discriminatory institution and should add new information not measured by other variables.
Statistical association
Variables included in the same dimension should be statistically associated, and thereby capture similar areas of social institutions without being redundant.
The SIGI includes 27 variables combined into 16 indicators and 4 dimensions (sub-indices). The 27 variables include:
-
1. 14 categorical variables describing the level of discrimination in legal frameworks (for all SIGI indicators but two: missing women and FGM) – these variables are based on 144 questions out of the 312 used to draft the SIGI country profiles;
-
2. 3 attitudinal variables describing the level of discrimination in social norms; and
-
3. 10 variables on prevalence rates describing the level of discrimination in practices.
The SIGI and its dimensions are constructed according to the steps below.
Step 1. Building the Gender, Institutions and Development Database
Truncating quantitative data at the equality benchmark and inverting the scale.
First quantitative data are collected and harmonised to be comparable across countries/territories. Data sources vary according to the country/territory and variable. For example, prevalence of girl child marriage is collected through the UN World Marriage Database, while women’s share of MPs is based on the Inter-parliamentary Union database.
The SIGI and its sub-indices range from 0 for no discrimination to 1 for absolute discrimination.
For some variables, equality is reached at 0.5 instead of 1. Equality in political representation for example is achieved when 50% of MPs are women. According to the variables, the scale from no discrimination to absolute discrimination may be inverted to fit with the 0-1 scale. Moreover, quantitative variables are truncated at the equality benchmark. Therefore, countries/territories having 50% or more of female members of parliament have a score of 0.
The SIGI country profiles contain fully referenced qualitative information relative to social institutions, organised by “dimensions” presented as sub-indices in this report. They were drafted following a standardised structure to ensure comparability across countries/territories in line with the following guidelines:
Conceptual relevance
Qualitative information should be relevant to the conceptual framework of discriminatory social institutions.
Sources
All information should be referenced and sourced from constitutions, legal frameworks, and primary publications, reports or studies, using the most recent data.
Data should be sourced from and cross-checked with reliable studies, reports and publications, including country reports to the Convention on the Elimination of All Forms of Discrimination against Women, reports by international organisations and country sources.
Validation
Country profiles were developed through a multiple-stage internal draft and review process. Qualitative information was validated by external gender experts with knowledge of the policy and legal landscape for gender equality and women’s rights at a national level (see www.genderindex.org).
Assigning a score to qualitative variables
The qualitative information detailed in the SIGI country profiles are quantified using the following coding manual:
0: The legal framework provides women with the same rights as men, without legal exceptions for some groups of women. There are no customary, traditional or religious laws or practices that discriminate against women.
0.25: The legal framework provides women with the same rights as men, without legal exceptions for some groups of women. However, some customary, traditional or religious laws or practices do discriminate against women.
0.5: The legal framework provides women with the same rights as men. However, it does not apply to all groups of women.
0.75: The legal framework restricts some women’s rights.
1: The legal framework fully discriminates against women’s rights.
In cases where no, or where insufficient information exists, variables are not assigned a value. The legal indicators are assessed based on all applicable legal frameworks, including civil law, religious law, customary law and traditional law.
Step 2: Constructing indicators
Some indicators are based on one variable while others on several. In the latter case, the indicator is calculated only if all variables are assigned a value. For example:
Violence against women=ln(1/3 e^(Laws on violence against women)+1/3 e^(Attitude towards domestic violence)+1/3 e^(Prevalence of domestic violence))
Step 3: Aggregating indicators to build the sub-indices
The dimensions aim to provide a summary measure of each area of discrimination. The dimension is calculated only if all indicators are assigned a value. For example:
Discrimination in the family=ln(1/4 e^(Child marriage)+1/4 e^(Household responsibilities)+1/4 e^(Inheritance)+1/4 e^(Divorce))
Step 4: Computing the SIGI
The SIGI is a composite indicator built as an unweighted average of a non-linear of the dimensions. The SIGI is calculated only if all sub-indices are assigned a value.
SIGI=ln(1/4 e^(Discrimination in the family)+1/4 e^(Restricted physical integrity)+1/4 e^(Restricted access to productive and financial resources)+1/4 e^(Restricted civil liberties))
Why exponentiate each sub-index and indicator?
-
As SIGI is a multidimensional index, the use of this formulation helps investigate the trade-offs between the sub-indices/indicators/variables.
-
These trade-offs, however, are partial: an increase in inequality in one sub-index /indicator/variable can only be substituted partially by a decrease in inequality in another sub-index/indicator/variable.
-
The magnitude of the increase of inequality is largest in the sub-index/indicator/variable where the country already performs poorly.
-
While the SIGI 2014 used the “quadratic mean” formulation to compute the level of discrimination, the SIGI 2019 uses exponential and logarithmic functions. This allows more variability in the rates of substitution between low levels and high levels of discrimination.
Why are the sub-indices/indicators/variables equally weighted?
-
Each sub-index/indicator/variable of discriminatory social institutions has equal value.
-
No sub-index/indicator/variable is more important than another in terms of deprivation experienced by women.
How are the SIGI categories defined?
The SIGI classification clusters 120 countries/territories into five levels of institutional social and statutory discrimination: very low, low, medium, high and very high.
Definition of variables
Notes
1 ActionAID (2013)
Bangladesh Bureau of Statistics (2013)
Cabo Verde National Institute of Statistics (2012)
ECLAC (2010)
ECLAC (2016)
General Statistics Office of Viet Nam (2014-15)
Instituto Nacional de Estadistica y Censos – Costa Rica
National Institute of Statistics of Rwanda (2013-14)
National Statistics, Republic of China (Chinese Taipei) (2000)
OECD SIGI Burkina Faso country study (2018)
OXFAM (2017)
République de Guinée (2002-03)
Statistics Sierra Leone (2003-04)
Timor-Leste Ministry of Finance (2007)
Troisième enquête Camerounaise auprès des ménages
UNSD time use data portal (2016)
Zimbabwe 2014 LFS
2 ActionAID (2013)
Bangladesh Bureau of Statistics (2013)
Cabo Verde National Institute of Statistics (2012)
ECLAC (2010)
ECLAC (2016)
General Statistics Office of Viet Nam (2014-15)
Instituto Nacional de Estadistica y Censos – Costa Rica
National Institute of Statistics of Rwanda (2013-14)
National Statistics, Republic of China (Chinese Taipei) (2000)
OECD SIGI Burkina Faso country study (2018)
OXFAM (2017)
République de Guinée (2002-03)
Statistics Sierra Leone (2003-04)
Timor-Leste Ministry of Finance (2007)
Troisième enquête Camerounaise auprès des ménages
UNSD time use data portal (2016)
Zimbabwe 2014 LFS
Questions used to calculate the SIGI scores
The SIGI country profiles are derived from a questionnaire on gender-related legal frameworks – both formal and traditional/cultural – in the countries and territories covered. The questionnaires are mainly divided into two groups – the legal framework (formal laws) and the de facto (customary, religious or traditional practices or laws). While most of the questions have “Yes” or “No” answers, there are two that have numeric values (i.e. the question on the legal age of marriage for women and men). The list of the questions is as follows.