THEORY → MODEL → TEST
A theoretical proposition becomes experimentally useful when its assumptions can be translated into a model and the model produces outputs that can be compared with observations.
EXPERIMENTAL PURPOSE
DEFINE
Define the hypothesis, variables, initial conditions and measurable quantities before evaluating the result.
COMPARE
Compare predicted model behavior with independently obtained observations or measurements.
CHALLENGE
Allow observations to disagree with the model. A meaningful test must leave room for a contrary result.
TESTABLE HYPOTHESIS
A hypothesis becomes testable when it leads to a sufficiently specific expectation that can be compared against an independently obtained result.
INPUTS & INITIAL CONDITIONS
The experimental setup must distinguish the quantities supplied to the model from the quantities obtained from observation.
INPUTS
Values supplied to the experimental or computational procedure.
INITIAL CONDITIONS
The defined state from which the model or experiment begins.
CONTROL VARIABLES
Variables that are held constant or deliberately controlled during a test.
PARAMETERS
Parameters define how the model behaves within the selected experimental configuration. Their values and units should be explicitly documented.
Changing a parameter may change the model output. Parameter sensitivity must therefore be distinguished from empirical evidence.
EXPECTED OUTPUT
The expected output is the result generated by the model before comparison with the relevant observation.
OBSERVED OUTPUT
An observed result must originate from the defined measurement or observation procedure rather than from the model itself.
This distinction prevents a model from being treated as its own evidence.
EXPECTED vs OBSERVED
| Layer | Source | Role |
|---|---|---|
| Expected | Model | Defines predicted behavior |
| Observed | Measurement / Observation | Provides comparison data |
| Difference | Comparison | Measures deviation |
| Interpretation | Analysis | Evaluates possible explanations |
ERROR & UNCERTAINTY
MEASUREMENT ERROR
Differences introduced by the measurement process, instruments or data acquisition procedure.
MODEL ERROR
Differences arising because the model is an approximation of the system being represented.
UNCERTAINTY
Limits on what can be concluded from finite, imperfect or incomplete information.
REPRODUCIBILITY
A test should document enough information for the procedure to be repeated under the same defined conditions.
ALTERNATIVE EXPLANATIONS
Agreement between a prediction and an observation does not automatically establish that the proposed mechanism caused the observed result.
ALTERNATIVE MODEL
Another model may produce a similar prediction from different assumptions.
COMMON MECHANISM
The observed behavior may have an explanation that does not require the proposed mechanism.
UNCONTROLLED VARIABLE
An unaccounted variable may influence the measured result.
NEGATIVE RESULTS
A result that does not match the expected model output is not something to be hidden. It is a defined experimental outcome that may constrain the model or expose an incorrect assumption.
The next step is to examine measurement quality, model assumptions, parameter selection, initial conditions and alternative explanations before drawing conclusions.
EVIDENCE STATUS
MODEL OUTPUT
A calculated result produced by the model. It is not, by itself, empirical evidence.
OBSERVATION
A result obtained through an observation or measurement procedure.
INTERPRETATION
A reasoned assessment of what the comparison may imply. It remains dependent on the available evidence and assumptions.
EXPERIMENTAL LIMITATIONS
Every experiment operates within a finite range of conditions, instruments, parameters and available data. Conclusions should remain within those tested boundaries.
TEST → RESULT → REVISION
Experimental results can be used to evaluate the model, identify limitations and determine whether assumptions or parameters require revision.
The purpose of this loop is not to force the model toward agreement, but to make its assumptions and predictions increasingly explicit and testable.
MODEL LAB v4
The Model Lab provides the computational interface for examining defined model behavior under selected inputs and parameters.
OPEN MODEL LAB v4 →Protocols & Simulators
Planned measurements, model simulators and application notes now have their own section, each clearly labelled so a protocol, a model output and an analogy are never confused.
OPEN PROTOCOLS & SIMULATORS →TEST IT.