Scientific models can be designed for different purposes. Three important functional categories are descriptive, explanatory, and predictive models. These categories help identify what a model is intended to accomplish and how its performance should be evaluated.
A descriptive model represents the observed structure, behavior, or pattern of a system. It may summarize measurements, organize relationships, or show how components are arranged. A map, a food web diagram, and a statistical summary of population characteristics are examples of descriptive representations. Such models help scientists understand what is observed without necessarily explaining the underlying mechanisms.
An explanatory model represents relationships and mechanisms that help account for why or how a phenomenon occurs. It often incorporates causal assumptions or theoretical principles. For example, a kinetic molecular model explains gas pressure through molecular collisions, while a model of enzyme action represents interactions between enzymes and substrates.
A predictive model uses known information and established relationships to estimate outcomes that have not yet been observed or measured. Weather forecasting models, planetary orbit calculations, and population projections are examples. Predictions may concern future events or unknown conditions within a system.
These categories can overlap. A single climate model may describe current atmospheric conditions, represent physical mechanisms, and predict future climate behavior. The classification depends on the primary purpose of the model and the way its results are evaluated.
Descriptive, Explanatory and Predictive Models
Descriptive, explanatory, and predictive models are classified according to the main purpose they serve. Descriptive models summarize or represent observed patterns, explanatory models investigate mechanisms and relationships, and predictive models estimate possible future or unobserved outcomes.
A general model output can be expressed as:
Y = f(X, P)
A descriptive model organizes observed values of X, an explanatory model represents relationships among X and P, and a predictive model uses the relationship f to estimate Y for specified inputs. This general expression does not define a particular scientific model.
Explanation
Variables
| Symbol | Meaning | Si Unit |
|---|---|---|
Y | Observed, explained, or predicted model output | Depends on the output quantity |
X | Input variables, observations, or system conditions | Depends on the variables |
P | Model parameters or explanatory factors | Depends on the parameters |
f | Relationship used to connect model inputs and outputs | Not applicable by itself |
Example or application
A worked example has not yet been added.
Definition
Descriptive models represent observed patterns or structures, explanatory models represent mechanisms and relationships that account for phenomena, and predictive models use available information and modeled relationships to estimate unknown or future outcomes.
Purpose
To explain how scientific models differ according to their descriptive, explanatory, and predictive functions and how each type contributes to scientific investigation.
Importance
Understanding these model purposes helps researchers select appropriate methods, interpret model results correctly, and avoid confusing the description of an observed pattern with an explanation of its cause or a prediction of future behavior.
Model Types
| Name | Main Purpose | Typical Features | Examples |
|---|---|---|---|
Descriptive models | To represent or summarize observed structures, patterns, and relationships. |
|
|
Explanatory models | To represent mechanisms and relationships that explain how or why a phenomenon occurs. |
|
|
Predictive models | To estimate unknown, unobserved, or future outcomes. |
|
|
Worked Examples
| Title | Problem | Explanation |
|---|---|---|
Population Research | How can three model types be used to study a population? | A descriptive model summarizes the current population size and age distribution. An explanatory model investigates factors such as birth rate, death rate, migration, and resource availability. A predictive model estimates future population size under specified assumptions about those factors. |
Weather Investigation | How can different models contribute to weather science? | A descriptive model presents current temperature, pressure, and wind patterns. An explanatory model represents atmospheric circulation and heat transfer. A predictive model uses observations and physical relationships to estimate future weather conditions. |
Evaluation Criteria
| Model Type | Criteria |
|---|---|
Descriptive | Accuracy of representation, completeness of relevant observations, clarity, and faithful summarization of patterns. |
Explanatory | Consistency with evidence, plausibility of mechanisms, explanatory scope, and ability to distinguish alternative explanations. |
Predictive | Accuracy of predictions, performance on independent data, uncertainty estimates, and reliability under intended conditions. |
Relationships
Descriptive models can identify patterns that motivate explanatory models.
Explanatory models can generate predictions that can be tested experimentally.
Predictive performance can provide evidence about the usefulness of an explanatory framework.
A single scientific model may perform all three functions when it is designed and evaluated appropriately.
Limitations
Descriptive patterns alone do not establish causal mechanisms.
Explanatory models may contain assumptions that are difficult to test directly.
Predictive models can perform poorly when conditions change beyond their development data or assumptions.
A model can describe observations accurately without predicting future outcomes reliably.
Prediction accuracy does not necessarily prove that every mechanism assumed by a model is correct.
Common Misconceptions
A descriptive model does not automatically explain why a pattern occurs.
An explanatory model is not necessarily capable of accurate long-term prediction.
A predictive model does not guarantee that its forecast will occur.
A single model may combine descriptive, explanatory, and predictive functions.
A model's purpose must be considered when evaluating its success.
Reference
National Academies of Sciences, Engineering, and Medicine — A Framework for K-12 Science Education.
OpenStax — introductory scientific modeling and statistical concepts.
Related Topic
SCI-REF-005-003
SCI-REF-005-006
SCI-REF-005-007
SCI-REF-005-013
Sources and verification
Sources are shown as provided in the project data. Not every source has yet been independently verified. Please report errors or missing citations.
- Source verification pending; check an authoritative textbook, university, standards body, or peer-reviewed source.