Gen Z Religious Switching Model

ISEF Category: Behavioral and Social Sciences

Subcategory: Sociology and Anthropology  ·  Difficulty: Intermediate  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

A small shift in one age group can change the religious map of a whole region. Gen Z often switches affiliation faster than older adults, so a model can show when "nones" move from minority to majority. You can turn survey data into a simulation and test where those tipping points appear.

What Is It?

An agent-based model is a computer model where each person follows simple rules. Think of it like a crowd simulation in a game, except each dot stands for a student or young adult who may keep, leave, or change a religious label. When you run many dots together, small choices can add up to a big shift.

Calibrated against Pew Religious Landscape data means you tune the rules until the model matches real survey patterns. Then you can test a tipping point, which is the moment when a trend speeds up and a "none" majority becomes possible in a region. The goal is not to guess one perfect future. The goal is to see which rules make the pattern rise, slow down, or stall.

Why This Is a Good Topic

This topic works well because you can test it with public survey data and simulation, not a wet lab. It connects to real questions about identity, community change, and regional culture. You can learn how to clean data, build rules from evidence, and judge whether a model matches real trends instead of just looking nice on a graph.

Research Questions

• How does the peer-influence strength change the year when "nones" become the largest group in a region?
• What is the effect of using different region-specific starting shares on the tipping point?
• Does adding a family-retention rule delay majority-none status more in some regions than others?
• To what extent do switching rates from Pew data predict the model's regional outcomes?
• Which combination of age, region, and switching probability best matches the observed Pew pattern?
• How does uncertainty in the survey weights change the range of possible tipping-point years?

Basic Materials

• Laptop with spreadsheet software.
• Python or R installed.
• Public Pew Religious Landscape Study tables.
• CSV file editor or spreadsheet import tool.
• Notebook for assumptions, parameters, and notes.
• Charting tool such as Google Sheets or Excel.

Advanced Materials

• Python with pandas, numpy, scipy, and mesa.
• RStudio with tidyverse, ggplot2, and lme4.
• Stata, SPSS, or another statistical package for survey-weighted checks.
• Regional demographic datasets for age and population shares.
• High-memory workstation for Monte Carlo runs and sensitivity tests.

Software & Tools

• Python: Runs the agent-based simulation and parameter sweeps.
• Jupyter Notebook: Keeps code, charts, and notes in one place.
• R: Fits weighted summaries and checks whether results hold.
• Tableau Public: Builds clear regional charts without paid software.

Experiment Steps

1. Define the exact switching outcome you will measure, such as affiliation change, disaffiliation, or majority-none status.
2. Choose the minimum set of rules your agents will follow, and write down why each rule comes from Pew patterns.
3. Build a baseline model that matches one region first, then compare the same rules across other regions.
4. Add controls for survey weight, cohort size, and starting affiliation share so the model does not overfit one data table.
5. Plan sensitivity tests that change one assumption at a time and record how much the tipping point moves.

Common Pitfalls

• Using national averages only, which hides regional tipping points.
• Treating every Gen Z student as identical, which wipes out the age and region differences the model needs.
• Fitting the model until it matches Pew tables exactly, which can make the rules meaningless.
• Comparing simulation output to raw counts instead of percentage shares, which breaks the calibration.
• Calling any fast change a tipping point, which confuses noise with a real threshold.

What Makes This Competitive

A stronger version of this project does more than match a chart. It compares at least two model structures, tests whether the same rules work across regions, and shows uncertainty with confidence bands or simulation ranges. You can make it even stronger by holding out part of the Pew data and checking whether the model still predicts the unseen part well.

Project Variations

• Compare regional tipping points in the South, Midwest, Northeast, and West instead of modeling the country as one group.
• Test whether family influence or peer influence explains the switching pattern better.
• Swap affiliation switching for attendance or self-identification strength and see which measure changes first.

Learn More

• Pew Research Center Religious Landscape Study: Look for the report tables and downloadable datasets on religion, switching, and age groups.
• Pew Research Center on Religious Switching: Search the site for articles about why people change affiliation, especially among younger adults.
• MIT OpenCourseWare: Search for classes on computational social science or modeling to learn how agent-based models work.
• Journal of Artificial Societies and Social Simulation: Read open-access papers on agent-based models and social behavior.
• U.S. Census Bureau: Use regional age and population tables to compare your model with real denominators.