The wording is from ECOSOC's 1997 agreed conclusions, which is why it reads like a legal text rather than a slogan. Every clause was negotiated, and each closes off a common evasion.

189 governments adopted the Platform for Action, and it remains the most comprehensive intergovernmental statement on women's rights. It is not a treaty and creates political rather than legal obligation.

A quick diagnostic on any proposal: delete the gender paragraph and see whether the rest of the document changes. If nothing else moves, gender was an annex.

None of these assumptions is written down anywhere. They are inherited from the designer's own experience of using services, which is why they survive review by people with similar experience.

The legal frame in India accommodates both: Article 14 guarantees equality, and Article 15(3) explicitly permits special provision for women and children, which is what makes reservation and targeting constitutional.

India ratified CEDAW in 1993 with declarations, and reports periodically to the CEDAW Committee. The Committee's concluding observations are public and specific, and are rarely used domestically as an advocacy tool.

Each myth is a way of containing the work: making it small, making it someone else's, or making it finished. That is worth naming out loud, because the response differs for each.

The last row is the whole reason the distinction matters for practitioners. If gendered outcomes were biological they would be fixed; because they are social, they are a legitimate object of policy.

Two of the three roles Moser identified are invisible to the national accounts, which is why time-use surveys matter: they are the only instrument that makes unpaid work countable.

The triple role means a new activity does not slot into free time, because there is little. It displaces something else — usually rest, or a daughter's schooling.

Programmes overwhelmingly deliver in the first column, because condition improvements are visible, measurable within a project cycle and uncontroversial.

The point of separating the three is that resources alone do not constitute empowerment. A woman may have an account she does not operate, a title to land she cannot sell, a vote cast by instruction.

Crenshaw's term names a specific analytical claim: the disadvantages are not additive, so a Dalit woman's experience is not the sum of a separate caste effect and a separate gender effect.

NALSA v. Union of India (2014) recognised transgender persons as a third gender with the right to self-identification, and the Transgender Persons Act followed in 2019 — with substantial criticism from trans-rights groups over its provisions.

The frameworks are not interchangeable, because they were built to answer different questions and carry different politics. Choosing one is already an analytical decision.

Its strength is that it is quick, concrete and produces something a sceptical manager will read: a table. Its limitation is that it describes the division of labour without asking why it exists.

Moser set out the framework in Gender Planning and Development (1993), as a planning methodology rather than a research tool — it was written to be used by the people designing programmes.

The SRA's contribution is to move the analysis off the individual woman and onto the rules, resources, people, activities and power in each institution — which is where the pattern is actually produced.

The four questions are the portable version of the frameworks. A practitioner who asks only these of every design will catch most of what a formal framework would.

The timing point is the one that decides everything else. An analysis commissioned after the design is approved can only annotate it, and everyone involved knows that.

The distinction matters because "we disaggregate by sex" is often given as a complete answer. It tells you the size of a gap in the things you already measure, and nothing about what you never asked.

Every wrong answer here treats the household as a single unit with one set of interests. Intra-household allocation research has shown for decades that it is not.

Every row is documented in the research literature on microfinance, self-help groups and financial inclusion in South Asia, and each was initially reported as a success on the access measure alone.

This is a few days of work, not a study. The reason it does not happen is rarely cost — it is that nobody is accountable for it and nothing fails without it.

The metaphor of leakage is exact. Gains made at one stage do not survive a gap at the next, and the commonest gap is between design and implementation.

None of these is a gender activity. They are ordinary design choices whose gendered effects are entirely predictable and rarely predicted.

Childcare is the intervention with the clearest evidence and the lowest adoption rate. Its absence is usually the single largest determinant of whether women can attend anything at all.

The household-level trap deserves emphasis: a benefit recorded against a household tells you nothing about who inside it received it, and most administrative data is recorded that way.

Markers are self-assigned at design and rarely audited, which is their known weakness — several evaluations of donor gender markers have found systematic over-scoring.

Sarah White's typology of participation makes a similar distinction, and the useful move for a practitioner is to decide in advance which level you are aiming at and how you will know.

The evidence on backlash is mixed and context-dependent: some studies find income gains reduce violence, others find increases where the gain challenges the man's provider role. Both findings are real and the design implication is the same — anticipate it.

The revenue half is nearly always omitted and is genuinely gendered: indirect taxes fall harder on households where women control small budgets, and exemptions on particular goods have gendered incidence.

Part B is where the classification judgement sits, and it is where critics have concentrated: a scheme reported at 30% for women is asserted, not measured, and the assertion is not verified.

Most of the spending that determines a woman's day is decided below the union budget, so a gender budget exercise confined to the centre misses the allocations that matter most.

The five steps come from Debbie Budlender and Rhonda Sharp's framework, developed from the South African and Australian experience, and they are sequenced deliberately: each depends on the one before.

Compare budget estimates, revised estimates and actuals for gender-tagged lines. A large gap between the first and the last is the most common and least discussed failure.

The bridge column is the point: most "unmeasurable" gender outcomes have a measurable proxy, and the reason they go unmeasured is that nobody specified one.

These are already collected, nationally and by district, which means a programme can benchmark against a real population figure rather than inventing a baseline.

These are the WHO ethical and safety recommendations for research on violence against women, and they are not optional refinements — they exist because the research itself can put a respondent at risk.

The Gender Gap Index measures gaps rather than levels by design, so a country with universally low education can score well on the education subindex. That is a feature, and it is routinely misread.

Sensitivity is the property most often missing. An indicator that will not change within the programme period cannot guide implementation, whatever else it does.

Most Significant Change and outcome harvesting are the usual named methods here, and both work by collecting accounts of change and then deciding collectively which matter.

The twin-track formulation is not a compromise between two camps. It is a response to twenty years of evidence that each approach fails in a specific and predictable way on its own.

Read the two columns as mirror images. Each track's failure is the other track's function, which is the actual argument for running both.

Reservation is the clearest case: no amount of mainstreaming would have placed over a million women in local government within a decade, which the 73rd Amendment did.

The sequencing logic is that Track 2 builds the capability to claim space, and Track 1 creates the space to claim. Doing the second without the first produces open doors nobody walks through.

The last row is the live risk. Engaging men can slide into men becoming the visible voice of gender work, which reproduces the pattern the work exists to change.

The final row is the real test and the hardest. Most organisations that describe themselves as having mainstreamed gender have in fact hired one committed person.

Procurement and partnership criteria are the most underused lever an organisation holds. Whatever you require of a grantee, they will produce; whatever you merely encourage, they will not.

Time is the shortfall nobody costs. A gender analysis takes days of a programme manager's time, and if that time is not in the workplan it comes out of something else or does not happen.

The pattern named in the last row is documented across the sector: the role is given to a junior woman as an addition to her existing job, with no budget and no authority.

The first three produce information. Only the fourth changes behaviour, and it is the one almost never implemented, because it requires a manager's appraisal to actually depend on it.

The ILO's participatory gender audit is the best-documented methodology, and its distinctive feature is that staff assess their own organisation rather than being assessed.

Each of these is measurable, which is the useful point: culture is often treated as intangible and therefore unmanageable, when much of it leaves a record.

Staff calibrate effort to what leadership actually asks about, not to what the policy says. One senior person asking the disaggregation question in every review does more than a training programme.

These are distinct diagnoses with distinct remedies, and they are often collapsed into a general complaint that mainstreaming has failed. Which one applies changes what you should do.

Longwe's term captures a specific mechanism rather than a general failure: each handover down the bureaucracy is a point at which the gender content can be lost without anyone deciding to drop it.

The form survives because it is what gets checked. A donor template asks for a gender paragraph, so a gender paragraph is produced, and the review verifies its presence rather than its consequence.

The critique — associated with Sylvia Walby, Shahra Razavi, Aruna Rao and others — is not that tools are bad. It is that the tools made the agenda adoptable by institutions precisely by removing what made it contentious.

The efficiency argument — that gender equality raises growth — is effective with finance ministries and carries a risk: an argument that rests on payoff can be defeated by evidence about payoff.

Every fix in the right column is cheap and structural. None requires a change of attitude, which is why they are the ones to push for first.

The imported-agenda claim is answerable on the facts: India's constitutional guarantees, its women's movement, and its own legislation on dowry, domestic violence and workplace harassment are domestic in origin.

The second is a genuine fork. An organisation that cannot fund gender work honestly is better off saying so than producing an unfunded policy that generates reports.

The machinery is dense by international standards and the outcomes remain uneven, which is the central puzzle of the Indian case and the reason implementation rather than legislation is usually the binding constraint.

The maternity amendment is the most instructive case for practitioners: a genuine expansion of entitlement whose cost falls on the employer, with an argued and contested effect on women's hiring.

Chattopadhyay and Duflo's study of reserved panchayat seats is the most-cited evidence that who holds the seat changes what gets built, and it has been extended and qualified by later work.

Read at least one critique alongside the frameworks. A practitioner who knows only the tools will apply them competently to the wrong question.

Each test is a question you can ask in a meeting without preparation, which is the point. Frameworks live in documents; these live in the conversation where a design is decided.