Historical Context & Motivation
Whenever people build a mathematical model — whether it predicts the trajectory of a rocket, the profit of a business, or the spread of a disease — they rely on input values that are often estimates or approximations. Sensitivity analysis is the practice of asking: "What happens to the output if one of those inputs changes?" This question has been central to applied mathematics for centuries, and it remains one of the most important skills in modeling today.
The roots of sensitivity reasoning go back to early scientists who wanted to understand how measurement errors could ripple through their calculations. As engineers began designing bridges, aircraft, and financial systems, the need for systematic methods grew dramatically. Today, sensitivity analysis is a standard tool in fields ranging from economics to climate science.
The central question sensitivity analysis addresses is deceptively simple: which inputs matter the most, and how much can they shift the output? As you learn to answer this question, you gain the ability to evaluate any model critically — deciding which assumptions to trust and which to question.
Core Principles & Definitions
Before diving into calculations, it helps to understand the core ideas that make sensitivity analysis work. These principles apply whether you are working with a simple formula or a complex simulation. Mastering them will give you a framework for thinking about any model.
Input vs. Output
One-at-a-Time (OAT) Testing
Sensitivity (Rate of Change)
Percentage Change for Fair Comparison
Critical Inputs
Visual Explanation
The diagram below shows a simple model: a lemonade stand where profit depends on price per cup and number of cups sold. We vary one input at a time by ±20% and watch how profit responds. The steeper the line, the more sensitive the output is to that input.
Notice that both lines pass through the same base-case point ($40 profit) in the center. As we move left or right — decreasing or increasing each input — the outputs diverge. The steeper the line, the more sensitive the output is. This type of graph is sometimes called a tornado diagram (when drawn horizontally with bars) or a spider plot (when drawn as lines radiating from a center point). Both are standard tools for visualizing sensitivity.
Mathematical Framework
To perform sensitivity analysis precisely, you need a few key formulas. These let you quantify exactly how much an output changes when an input shifts. You do not need calculus for these — just careful arithmetic and algebra.
Types of Sensitivity Analysis
There are several ways to approach sensitivity analysis depending on the complexity of your model and the question you are trying to answer. At the high-school level, two methods dominate: one-at-a-time (OAT) analysis and scenario analysis. The diagram below illustrates how they differ.
| Feature | OAT Analysis | Scenario Analysis |
|---|---|---|
| Inputs changed | One at a time | Several at once |
| Main strength | Isolates individual effects clearly | Shows realistic combined effects |
| Main limitation | Ignores interactions between inputs | Harder to pinpoint which input drives the change |
| Best used when | You want to rank inputs by importance | You want to plan for best/worst outcomes |
Worked Example
A school club sells custom t-shirts to raise money. Their profit model is:
The club wants to know: Is their profit more sensitive to selling price or to the number of shirts sold? We will use a 10% increase in each input to find out.
Strengths & Limitations
Like any analytical tool, sensitivity analysis has clear strengths and important limitations. Understanding both will help you use it effectively and recognize when additional techniques might be needed.
| Strengths | Limitations |
|---|---|
| Identifies which inputs have the greatest impact on the output, helping you prioritize where to gather better data. | OAT analysis assumes inputs change independently, but in reality inputs can be correlated (e.g., price and demand). |
| Requires only basic algebra — accessible without advanced math. | Provides no information about the probability of an input actually changing by a given amount. |
| Helps communicate risk and uncertainty to decision-makers in a clear, visual way. | For non-linear models, results can change depending on the size and direction of the input change. |
| Works with any model — formulas, simulations, or even mental models. | Does not guarantee that the model itself is correct — garbage in, garbage out. |
Connection to Advanced Theory
The sensitivity reasoning you are learning now is the foundation for more powerful techniques you will encounter in college-level courses. Calculus, statistics, and data science all build on this same core idea: understanding how changes in inputs propagate through a system to affect outputs.
| What You Learn Now | What Comes Next |
|---|---|
| ΔOutput ÷ ΔInput (average rate of change) | Partial derivatives (∂f/∂x) — the instantaneous rate of change of a function with respect to one variable |
| Sensitivity ratio using percent changes | Elasticity in economics — measures the responsiveness of demand to price, for instance |
| Scenario analysis (best/worst/base) | Monte Carlo simulation — running thousands of random scenarios to map out a probability distribution of outcomes |
| Ranking inputs by impact | Global sensitivity analysis (Sobol indices) — advanced methods that account for interactions between inputs |
If you take AP Calculus, you will learn that the derivative of a function is essentially the limit of the sensitivity ratio as the input change approaches zero. In AP Statistics or college econometrics, you will encounter regression coefficients that quantify sensitivity in data-driven models. The intuition you build now — asking "what if this changes?" — is the same mindset that drives advanced research across every STEM field.
Practice Problems
Lesson Summary
Sensitivity analysis is the systematic study of how changes in a model's inputs affect its outputs. The core tools include calculating the absolute change (ΔOutput), the percent change for fair comparison across different units, and the sensitivity ratio (percent change in output ÷ percent change in input) to determine whether an output is amplified or dampened relative to a given input.
Two key approaches are one-at-a-time (OAT) analysis, which isolates each input's effect, and scenario analysis, which examines best-case, base-case, and worst-case combinations. Inputs with the highest sensitivity ratios are critical inputs — they deserve the most attention when you are refining estimates or managing risk. While sensitivity analysis is powerful, remember that it does not tell you the probability of a change occurring, and results for non-linear models depend on the size and direction of the change tested.