Why Validating Your Model Matters: Ensuring Reliable Results

1 Why Test Model Results?

Results that can withstand validation are more convincing. After turning a problem into mathematics, building a model, solving it, and interpreting the outcome, one must not forget to validate the model. Whether this step is performed often reveals whether the modeler has mastered mathematical modeling.

2 Testing Model Parameters

During model construction and solution we try to choose reasonable parameters—for example, when setting an investment budget we consider interest rates within a range of a%‑b% and use a specific value c for calculations. Since the future is uncertain, the parameter may vary within that range. We care not only about the result under the chosen parameters but also about how the result changes if the parameters vary.

If adjusting a parameter (e.g., replacing c with another plausible value) leads to little change in the outcome, the result is said to be “insensitive” to that parameter, and we can confidently use the result for decision‑making. Conversely, if the outcome changes dramatically, the result is “sensitive,” indicating that the parameter is critical and must be estimated as accurately as possible for the result to be reliable.

Sensitivity analysis—examining how parameter variations affect results—is a common validation method. Besides parameters, other modeling choices (such as data normalization methods) can be varied. By re‑running the model with alternative processing methods and comparing outcomes, we further assess robustness. Consistent results across methods suggest reliability; large discrepancies warrant deeper investigation.

3 Checking Results Against Common Sense or Domain Knowledge

Model outputs are used to expand knowledge or guide practice. If a result clearly contradicts reality, it cannot be accepted. Different people may have different notions of “common sense” versus “professional knowledge.” When the modeler lacks sufficient domain expertise, validation can be performed by consulting relevant literature, searching for data, or discussing with knowledgeable peers.

4 Using Validation to Extend Results

Beyond confirming reasonableness, validation can broaden results. It is not merely a checklist of rigor; it introduces alternative perspectives—different methods or facts—that evaluate the outcome from new angles. These new viewpoints add content, supplementing and extending the original model.

In summary:

Model results become more grounded through validation.

The modeling process becomes more complete with validation.

The modeling mindset becomes more comprehensive through validation.