Data Feminism 101 - Workshop 2

75-90 Minutes

Workshop 2: Hands-On Analysis & Communication

Target Audience: Data practitioners ready to apply feminist principles to real datasets and visualizations

Prerequisites: Workshop 1 or familiarity with data feminism concepts, basic data analysis skills

Materials Needed: Computers with R/Python/Excel, sample datasets, presentation materials

Learning Objectives

By the end of this workshop, participants will be able to:

Conduct intersectional data analysis using practical techniques
Create data visualizations that challenge rather than reinforce bias
Design research questions through feminist and equity lenses
Communicate data findings in ways that center marginalized voices
Evaluate and improve existing data products using feminist frameworks

Part 1: Intersectional Analysis in Practice

25 minutes

Moving Beyond Single-Variable Analysis

Case Study: Analyzing Digital Financial Inclusion

Standard Analysis: "65% of women have bank accounts vs 78% of men - there's a gender gap in financial inclusion."

Intersectional Analysis Reveals:

Urban upper-caste women: 89% banked
Rural Dalit women: 34% banked
Disabled women (all categories): 23% banked
Transgender individuals: 12% banked
Single mothers under 25: 28% banked

Key Insight: "Gender gap" masks enormous variation - the real story is about intersecting systems of exclusion.

Practical Intersectional Analysis Methods

Multi-Way Cross-Tabulation

Break down outcomes by multiple identity categories simultaneously

Tool: Pivot tables, dplyr group_by(), pandas groupby()

Disaggregated Regression

Include interaction terms between identity variables

Tool: lm(outcome ~ gender*caste*location) in R

Subgroup Analysis

Separate models for different populations

Tool: Filter datasets by identity groups

Vulnerability Mapping

Create composite vulnerability scores across dimensions

Tool: Weighted indices, factor analysis

Hands-On Analysis Exercise (15 minutes)

Dataset: NFHS-5 Women's Empowerment Module (provided)

Research Question: How does women's decision-making autonomy vary across different social groups?

Your Task:

Single-dimension analysis (3 min): Calculate % women with decision-making autonomy by education level
Two-way analysis (5 min): Break down by education + wealth quintile
Intersectional analysis (7 min): Add caste/tribe + residence + working status

Questions to Explore:

Which group has highest autonomy? Lowest?
Does education have same effect across all castes?
How do urban/rural patterns differ by social group?
What story emerges that the simple education analysis missed?

# Sample R code starter:
library(dplyr)
# Simple analysis
women %>% group_by(education) %>% 
  summarise(autonomy_pct = mean(decision_autonomy, na.rm=T))

# Intersectional analysis  
women %>% group_by(education, wealth, caste, urban) %>%
  summarise(autonomy_pct = mean(decision_autonomy, na.rm=T)) %>%
  arrange(autonomy_pct)
                

Part 2: Feminist Data Visualization

20 minutes

Challenging Visual Bias

Problematic Approach

Chart Title: "India's Growing Economy"

Visual: Bar chart showing rising GDP per capita

Problems:

Hides inequality within growth
No disaggregation by social groups
Implies universal benefit
Masculine framing ("growing")

Feminist Alternative

Chart Title: "Who Benefits from Economic Growth?"

Visual: Small multiples showing income growth by gender, caste, region

Improvements:

Centers equity questions
Shows distribution, not just averages
Makes invisible groups visible
Invites critical analysis

Feminist Visualization Principles

Do:

Disaggregate ruthlessly: Show patterns within patterns
Use inclusive color palettes: Avoid gendered pink/blue defaults
Include uncertainty: Show confidence intervals, missing data
Provide context: Historical trends, policy changes, external factors
Center margins: Highlight most vulnerable groups prominently

Avoid:

False universalism: "Indians prefer..." when you mean "urban English-speakers prefer..."
Deficit framing: Always showing marginalized groups as "lacking"
Binary thinking: Male/female charts when identity is more complex
Decontextualized data: Numbers without historical/political context

Part 3: Problem Set - Applied Feminist Data Analysis

25 minutes

Problem Set: Education Access Analysis

Context: You're analyzing primary school enrollment data for a state education department. The headline finding is "96% primary enrollment achieved - nearly universal access!"

Dataset Provided: District-level enrollment by gender, caste, disability status, migration status

Problem 1: Intersectional Enrollment Analysis (10 minutes)

Task: Using the provided data, conduct an intersectional analysis of enrollment.

Questions to Answer:

What is the enrollment rate for SC girls vs. general category boys?
How do migration patterns affect enrollment differently for different castes?
Which intersectional group has the lowest enrollment? Highest?
Calculate the "equity gap" - difference between highest and lowest groups

# Data structure:
# district | gender | caste | disability | migrant | enrolled | total
# Calculate intersectional enrollment rates
enrollment_rates <- data %>%
  group_by(gender, caste, disability, migrant) %>%
  summarise(
    enrollment_rate = sum(enrolled)/sum(total),
    sample_size = sum(total)
  ) %>%
  arrange(enrollment_rate)
                

Problem 2: Visualization Redesign (8 minutes)

Current Viz: Single bar chart showing "96% enrollment rate"

Your Task: Design a feminist alternative visualization

Requirements:

Show intersectional breakdowns
Use inclusive design principles
Include uncertainty/sample sizes
Write equity-centered title and caption

Problem 3: Policy Memo (7 minutes)

Task: Write a 200-word policy memo based on your analysis

Structure:

Opening: Challenge the "universal access" narrative
Evidence: Key intersectional findings
Recommendations: Targeted interventions for most excluded groups
Conclusion: Reframe success metrics around equity

Part 4: Research Design Through Feminist Lens

15 minutes

Designing Equitable Research Questions

Traditional Question	Feminist Reframe	Why It Matters
"What factors influence women's labor force participation?"	"How do care responsibilities, safety concerns, and social norms create different barriers for different women's economic participation?"	Recognizes diversity among women; centers structural barriers
"Are microfinance programs effective?"	"How do microfinance programs affect different women's autonomy, with attention to caste, class, and household dynamics?"	Moves beyond simple effectiveness to examine differential impacts
"What is the digital divide in India?"	"How do intersecting identities shape access to and meaningful use of digital technologies?"	Centers intersectionality; examines quality not just access

Research Question Makeover (10 minutes)

Step 1 (2 min): Choose a research question from your current work

Step 2 (5 min): Apply feminist reframing using these prompts:

How does this question assume universal experience?
Whose voices/experiences does it center? Who's missing?
What power dynamics does it ignore or take for granted?
How could intersectionality change what we ask?
What would marginalized communities want to know about this topic?

Step 3 (3 min): Rewrite your question using feminist principles

Partner Share: Compare original and reframed versions

Reflection & Action Planning

10 minutes

Personal Action Planning

Individual Reflection (7 minutes):

Commit to Change: Choose one concrete way you'll apply feminist data practice in your work over the next month:

Analysis: Add intersectional breakdown to routine report
Visualization: Redesign one chart using equity principles
Research: Reframe research question through justice lens
Communication: Change how you present findings to center margins
Data Collection: Advocate for better representation in sampling

Accountability: Write down your commitment and one person you'll share it with

Group Share (3 minutes): Quick round of commitments

Key Takeaway

Feminist data practice is not just about adding "gender" as a variable. It's about fundamentally changing how we ask questions, collect data, conduct analysis, and communicate findings to advance justice rather than reproduce inequality.

Technical Resources

R Packages for Feminist Analysis:

intersectionality: Tools for intersectional analysis
viridis: Inclusive color palettes
ggplot2 + facet_wrap: Small multiples for disaggregation
survey: Proper handling of complex survey data

Python Tools:

pandas.groupby(): Multi-level disaggregation
seaborn.FacetGrid: Intersectional visualization
plotly: Interactive dashboards for exploration
scipy.stats: Statistical testing across groups

Data Sources for Practice:

NFHS-5: Rich intersectional demographics
PLFS: Employment data with social categories
ASER: Learning outcomes with detailed breakdowns
Census 2011: Comprehensive population data

Next Steps in ImpactMojo:

Multivariate Analysis 101: Advanced statistical methods for intersectional data
Visual Ethnography 101: Non-quantitative ways of knowing
Care Economy 101: Measuring unpaid work and time use
Marginalized Identities 101: Data representation and identity

This handout is part of the ImpactMojo 101 Knowledge Series
Licensed under CC BY-NC-ND 4.0 • Free to use with attribution • www.impactmojo.in

For complete datasets, R/Python code, and analysis templates from this workshop, visit the ImpactMojo platform.