Subject Pack . S3 . Interactive

Gender Impact Assessment Design

Move beyond sex-disaggregation to substantive gender impact assessment. Choose frameworks, operationalise intersectionality, and design reporting that does not flatten gender to a checkbox. Walk out with a GIA design brief.

4 modules ~3 hours Interactive India-context
Your progress
0% complete
Your Capstone

Drafted GIA Design Brief

Walk in with a programme. Walk out with a gender impact assessment design brief -- framework selection with justification, intersectionality operationalisation, measurement plan, and reporting structure. Built automatically from your module answers.

Module 1 . ~25 min

Beyond sex-disaggregation -- what GIA actually measures

Sex-disaggregated data tells you how many women participated. Gender impact assessment tells you whether the programme changed gender relations. These are fundamentally different questions, and most evaluations stop at the first one.

The three layers of gender analysis

Why Indian evaluations stall at Layer 1

Three structural reasons:

  1. Funder reporting requirements focus on sex-disaggregated numbers because they are easy to report and compare. NITI Aayog's Aspirational Districts programme, for instance, tracks women beneficiaries as a count, not as an agency measure.
  2. Time horizons -- agency and norm change take 3-5 years to manifest. Most evaluations are commissioned at 18 months.
  3. Measurement difficulty -- validated instruments for agency and norms in Indian contexts are scarce. The WEAI was designed for Bangladesh and adapted for South Asia, but it measures agricultural empowerment specifically, not general agency.
Worked example

Mahila Samakhya ran women's collectives in Uttar Pradesh for over 25 years. External evaluations consistently reported high participation numbers. But ICRW's 2018 deep-dive found that participation alone did not predict changes in intra-household decision-making. What mattered was the duration of membership (3+ years) and whether the collective engaged in public action (gram sabha participation, PDS monitoring). The evaluation had to measure collective-level and household-level agency separately.

Your GIA Scoping Sheet

Fill these for your programme. Your answers save automatically and flow into the final capstone.

e.g., "NRLM-linked SHG livelihoods programme, 200 SHGs, 3 districts in Jharkhand"
Is this an 18-month midterm or a 5-year endline? This determines which layers are measurable.
Saved
Self-check
A funder asks you to evaluate the "gender impact" of a livelihoods programme. Their reporting template has one column: "% women beneficiaries." What is the most useful first response?
Fill the column and submit
Refuse the template and design a full GIA
Fill the column AND propose 2-3 additional indicators measuring agency or decision-making change
Explain that gender impact cannot be measured in one number
Correct. Meet the funder where they are (fill the column), then expand the frame. Proposing 2-3 concrete, measurable agency indicators alongside participation data is more persuasive than a lecture on gender theory.
Module 2 . ~30 min

Framework choice: WEAI vs Pro-WEAI vs GEM vs custom

Choosing a GIA framework is not an academic exercise. The framework determines what you measure, how long fieldwork takes, what skills your enumerators need, and what kind of findings you can report. Choose wrong and you spend Rs 20 lakh measuring the wrong thing.

Framework comparison

FrameworkMeasuresBest forIndia validationFieldwork burden
WEAI5 domains of agricultural empowerment + Gender Parity IndexAgriculture/livelihoods with explicit women's empowerment theoryStrong (IFPRI South Asia)High: 45-60 min survey per couple
Pro-WEAI12 indicators across 3 domains; adds intrinsic agency, self-efficacyProgrammes addressing psychological empowerment alongside economicModerate (piloted in Odisha, Maharashtra)Very high: 60-90 min per respondent
GEM ScaleGender-equitable attitudes on 24 itemsBCC programmes, adolescent programmes, norm-change interventionsGood (adapted by ICRW India, Population Council)Low: 15 min scale
Custom compositeProgramme-specific dimensions drawn from validated sub-scalesWhen no single framework matches your ToCDepends on sub-scale selectionVariable

Decision rules

The adaptation trap

Adapting a framework is not picking the items you like. It means testing whether the items measure the same construct in your context. The WEAI "input in productive decisions" item assumes women make agricultural decisions. In Jharkhand's mining-affected tribal communities, the relevant decisions may be about forest produce or wage labour instead. Cognitive pretesting in your specific context is mandatory, not optional.

Your Framework Selection

Choose and justify. These flow into your capstone.

e.g., "Qualitative life-history interviews with 30 women" or "GEM Scale alongside WEAI"
Saved
Self-check
You are evaluating an adolescent girls' life-skills programme in Rajasthan. The programme teaches negotiation, menstrual health, and financial literacy. Which framework is most appropriate as your primary measure?
WEAI -- measures agricultural empowerment across 5 domains
GEM Scale -- measures gender-equitable attitudes, adapted for adolescents by Population Council India
Pro-WEAI -- 60-90 min survey designed for adult women in agricultural households
NFHS household decision-making module
Correct. GEM Scale measures exactly what this programme targets: gender-equitable attitudes. It has been adapted for Indian adolescents by Population Council. WEAI and Pro-WEAI are designed for agricultural contexts and adult women; NFHS decision-making items are for married women.
Module 3 . ~30 min

Intersectionality in practice

Gender does not operate in isolation. A Dalit woman in rural Bihar, an Adivasi woman in Jharkhand's mining belt, and an upper-caste woman in urban Pune experience the same livelihood programme differently. Intersectionality is not a theoretical add-on -- it is a design requirement.

The practical problem

Intersectional analysis requires sufficient sample sizes across intersecting categories. If your sample is 300 women, and you want to disaggregate by caste (SC/ST/OBC/General) and geography (rural/urban), you need 300 / 8 = ~37 per cell. With 20% attrition, you need 45 per cell. That is 360 women minimum. Most GIA budgets do not account for this.

Three operational approaches

Indian intersections to consider

AxisWhy it matters for GIAData source
CasteSC/ST women face compounded discrimination in credit access, labour markets, and political participationSelf-reported; verify with community records
ReligionMuslim women in India have lower labour force participation (PLFS 2022-23: 15.6% vs 32.8% Hindu women)Self-reported
DisabilityWomen with disabilities are systematically excluded from most programmes; 2011 Census undercounts significantlyWashington Group Short Set questions
Age/marital statusAdolescent girls, young married women, and widows face distinct constraintsDemographic section
GeographyUrban/rural; plain/hill/tribal/coastal -- distinct livelihood structures and gender normsSampling frame
The ethics of naming

Intersectional analysis can identify marginalised sub-groups in small communities. In a 500-person village, reporting that "SC women who are also single mothers experienced increased harassment" can identify individuals. Aggregate to the block or district level for sensitive findings. Discuss this in your ethics protocol explicitly.

Your Intersectionality Plan

Design the intersectional analysis. These flow into your capstone.

e.g., "caste (SC/ST/OBC/General), geography (rural/peri-urban), age group (18-25, 26-40, 40+)"
Will you interview male household members? Use couple-level analysis?
Saved
Self-check
Your GIA sample is 200 women from 3 districts. The programme team wants disaggregation by caste (4 categories), geography (rural/urban), and disability status. How many cells does this create, and is 200 sufficient?
8 cells; 200 is sufficient (25 per cell)
16 cells; 200 is far too small (12.5 per cell) -- reduce axes or increase sample
12 cells; 200 is borderline
4 cells; 200 is plenty
Correct. 4 caste categories x 2 geography x 2 disability = 16 cells. At 200, that is 12.5 per cell -- far below the minimum 30 needed for quantitative analysis. Either increase the sample to 500+, drop an axis, or use qualitative methods for the thinnest cells.
Module 4 . ~25 min

Reporting that does not flatten gender to a checkbox

The most rigorous GIA is wasted if the report reduces findings to "the programme benefited women." Three reporting failures are endemic.

Failure 1: The aggregation trap

Reporting average effects hides the distribution. If empowerment scores improved for OBC women (+0.3 SD) but declined for SC women (-0.1 SD), the average effect (+0.15 SD) is misleading. Always report sub-group effects for your pre-specified intersections.

Failure 2: The binary framing

"The programme is gender-positive" or "gender-negative" is never the right frame. Programmes are simultaneously empowering on some dimensions and reinforcing on others. A microfinance programme may improve women's income (empowering) while increasing their workload without reducing domestic labour (reinforcing).

Failure 3: Disconnecting quant from qual

The most powerful GIA reports use qualitative data to explain quantitative patterns. If the WEAI score improved in one domain but not another, the life-history interviews explain why. Report them together, not in separate chapters.

Reporting structure for GIA

  1. 1-page executive summary with gender-specific headline findings and one recommendation per finding
  2. Gender context section -- baseline gender landscape, norms, constraints. This grounds the findings.
  3. Findings by domain (not by method) -- for each empowerment domain, present quant + qual together
  4. Intersectional findings -- dedicated sub-section showing how effects vary across intersections
  5. Unintended effects -- did the programme create new burdens, risks, or backlash for women?
  6. Recommendations -- actionable, specific, addressed to named stakeholders
Worked example

PRADAN's SHG programme in Jharkhand: The GIA report found that economic empowerment (measured by WEAI production domain) improved significantly. But qualitative interviews revealed that women's increased farm work came with no reduction in domestic labour -- they were working 14-hour days. The report framed this as "economic gains with time-poverty costs" and recommended integrating time-use analysis into future evaluations. The funder changed the programme design to include household-level negotiation sessions.

Your Reporting Plan

Design the report structure. These flow into your capstone.

e.g., increased workload, backlash, time poverty, new dependencies
"This GIA will tell us ___ and will NOT tell us ___."
Saved
Self-check
Your GIA finds: WEAI score improved on average (+0.2 SD). But disaggregation shows SC women declined (-0.05 SD) while OBC women gained (+0.35 SD). How should the executive summary frame this?
"Programme improved women's empowerment (effect size: +0.2 SD)"
"Results inconclusive due to variation"
"Empowerment gains concentrated among OBC women; SC women saw no improvement -- programme design needs caste-differentiated strategies"
"Average effect positive but not statistically significant for all sub-groups"
Correct. The honest frame names who benefited, who did not, and what the programme should do about it. Averaging over the distribution hides the most important finding -- that the programme may be widening caste gaps in empowerment.
Capstone

Your GIA Design Brief

You have completed the four modules. Click Build my brief to compile everything into a single GIA design brief.

Gender Impact Assessment Design Brief

Click "Build my brief" -- your module answers will be pulled into the artefact.

Your brief will appear here when you click "Build my brief".
Done?

Share this brief with a gender specialist before circulating. The most common blind spot is unstated assumptions about household structure.

Help us improve this pack: feedback form.

All Practice Packs →