# Hypothesis - Scientific Method for Problem Solving
## What It Does
Hypothesis applies the scientific method to questions and problems. By structuring observations into testable hypotheses with explicit variables and predictions, you move from guessing to systematic investigation.
**Key Value**: Stop guessing and start testing. Hypothesis forces you to make falsifiable predictions, design experiments, and update beliefs based on evidence.
## When to Use This Tool
- **Product decisions** - Test assumptions before building features
- **Performance issues** - Form hypotheses about bottlenecks
- **A/B testing design** - Structure experiments properly
- **Root cause analysis** - Investigate systematically
- **Market research** - Test assumptions about user behavior
- **Technical investigations** - Diagnose complex system issues
## How It Works
Hypothesis guides you through the scientific method:
```
Observation → Question → Hypothesis → Experiment → Analysis → Conclusion
```
### Scientific Stages
| Stage | Purpose |
|-------|---------|
| `observation` | Notice something interesting or problematic |
| `question` | Formulate what you want to understand |
| `hypothesis` | Propose an explanation with testable predictions |
| `experiment` | Design and run the test |
| `analysis` | Evaluate results against predictions |
| `conclusion` | Update understanding based on evidence |
| `iteration` | Refine hypothesis based on findings |
### Variable Types
| Type | What It Means |
|------|---------------|
| `independent` | What you manipulate/change |
| `dependent` | What you measure |
| `controlled` | What you keep constant |
| `confounding` | What might skew results |
## Interactive Examples
### Scenario 1: Investigating Conversion Drop
**Context**: Checkout conversion rate dropped from 3.2% to 2.1% after a website redesign.
**Input**:
```json
{
"stage": "hypothesis",
"inquiryId": "conversion-rate-study",
"iteration": 1,
"nextStageNeeded": true,
"observation": "Our checkout conversion rate dropped from 3.2% to 2.1% after the website redesign last month",
"question": "What specific design changes are causing users to abandon checkout?",
"hypothesis": {
"statement": "The new multi-step checkout process increases cognitive load and causes abandonment compared to the previous single-page checkout",
"variables": [
{"name": "checkout_steps", "type": "independent", "operationalization": "Number of pages in checkout flow (1 vs 4)"},
{"name": "conversion_rate", "type": "dependent", "operationalization": "Percentage of cart additions that complete purchase"},
{"name": "device_type", "type": "controlled", "operationalization": "Desktop vs mobile users analyzed separately"}
],
"assumptions": [
"Users prefer fewer steps",
"The product mix hasn't changed significantly"
],
"hypothesisId": "H1-checkout-steps",
"confidence": 0.75,
"domain": "UX/Conversion Optimization",
"iteration": 1,
"status": "proposed"
}
}
```
**Output**:
```json
{
"stage": "hypothesis",
"inquiryId": "conversion-rate-study",
"iteration": 1,
"nextStageNeeded": true,
"status": "success"
}
```
**What This Means**: You've formulated a testable hypothesis with:
- **Clear statement**: Multi-step checkout → lower conversion
- **Variables defined**: Steps (independent), conversion (dependent), device (controlled)
- **Assumptions stated**: Preferences about steps, stable product mix
- **Confidence calibrated**: 75% - you believe this but aren't certain
---
### Scenario 2: Designing the Experiment
**Context**: Moving from hypothesis to experiment design.
**Input**:
```json
{
"stage": "experiment",
"inquiryId": "conversion-rate-study",
"iteration": 2,
"nextStageNeeded": true,
"hypothesis": {
"statement": "The new multi-step checkout process increases cognitive load and causes abandonment",
"hypothesisId": "H1-checkout-steps",
"status": "testing"
},
"experiment": {
"design": "A/B test with 50/50 traffic split between old single-page and new multi-step checkout",
"methodology": "Randomized controlled trial with 2-week duration",
"predictions": [
{"if": "Multi-step checkout has higher abandonment", "then": "Conversion rate will be 30%+ lower in multi-step group", "else": "The issue is elsewhere in the redesign"},
{"if": "Cognitive load is the cause", "then": "Time-on-page before abandonment will be higher in multi-step", "else": "Users are abandoning for other reasons"}
],
"experimentId": "exp-checkout-001",
"hypothesisId": "H1-checkout-steps",
"controlMeasures": [
"Exclude users who arrived via specific promotional campaigns",
"Run during non-holiday period",
"Ensure equal distribution across device types"
]
}
}
```
**What This Means**: Your experiment is designed with:
- **Clear methodology**: Randomized A/B test
- **Explicit predictions**: If-then-else structure forces falsifiability
- **Control measures**: Reduce confounding factors
- **Success criteria**: 30% difference threshold is specific and measurable
---
### Scenario 3: Analyzing Results and Concluding
**Context**: Experiment complete, analyzing results.
**Input**:
```json
{
"stage": "conclusion",
"inquiryId": "conversion-rate-study",
"iteration": 3,
"nextStageNeeded": false,
"experiment": {
"experimentId": "exp-checkout-001",
"hypothesisId": "H1-checkout-steps",
"results": "Single-page conversion: 3.1%, Multi-step conversion: 2.0%. Difference: 35%. Time-on-page before abandonment: 45s (single) vs 89s (multi).",
"outcomeMatched": true,
"unexpectedObservations": [
"Mobile users showed even larger difference (40% lower conversion)",
"Abandonment peaked at step 2 (shipping details) not payment"
],
"limitations": [
"Could not control for user acquisition source changes",
"2-week period may miss weekly patterns"
],
"nextSteps": [
"Focus optimization on shipping details step",
"Consider progressive disclosure for mobile users"
]
},
"analysis": "Results support the hypothesis. The 35% conversion difference exceeds our 30% threshold. Extended time-on-page suggests cognitive load as the mechanism.",
"conclusion": "Multi-step checkout is causing the conversion drop. The shipping details step is the primary friction point. Recommend reverting to single-page or simplifying the multi-step flow, with priority on mobile experience."
}
```
**What This Means**: Your investigation reached a clear conclusion:
- **Hypothesis supported**: 35% difference confirmed the prediction
- **Mechanism identified**: Time-on-page data supports cognitive load theory
- **Unexpected findings**: Mobile worse, shipping step is the culprit
- **Actionable next steps**: Specific improvements recommended
## User Experience
Hypothesis produces rigorous investigation records:
| Field | What It Tells You |
|-------|-------------------|
| `stage` | Current phase of scientific inquiry |
| `inquiryId` | Unique identifier for this investigation |
| `iteration` | How many cycles of refinement |
| `nextStageNeeded` | Whether to continue investigating |
| `hypothesis` | Proposed explanation with variables |
| `experiment` | Test design with predictions |
| `analysis` | Evaluation of results |
| `conclusion` | Final understanding and recommendations |
**Investigation Flow**:
```
Observation → Question → Hypothesis → Experiment → Analysis → Conclusion
↑ ↓
└──── Iteration (refine) ──┘
```
## Integration Tips
- **Make hypotheses falsifiable** - If nothing could disprove it, it's not scientific
- **Define success criteria upfront** - Avoid post-hoc rationalization
- **Document unexpected findings** - Often more valuable than expected results
- **List limitations honestly** - Helps others evaluate your conclusions
- **Chain with Reflect** - Assess confidence in your hypothesis
- **Chain with Debug** - Use debugging approaches to find observations
## Quick Reference
| Parameter | Type | Required | Description |
|-----------|------|----------|-------------|
| `stage` | enum | Yes | observation, question, hypothesis, experiment, analysis, conclusion, iteration |
| `inquiryId` | string | Yes | Unique identifier for this inquiry |
| `iteration` | number | Yes | Current iteration count |
| `nextStageNeeded` | boolean | Yes | Whether to continue |
| `observation` | string | No | What you noticed |
| `question` | string | No | What you want to understand |
| `hypothesis` | object | No | Statement, variables, assumptions, confidence, status |
| `experiment` | object | No | Design, methodology, predictions, control measures, results |
| `analysis` | string | No | Evaluation of results |
| `conclusion` | string | No | Final understanding |
