S2.3 Concluding and evaluating

A conclusion is a bounded answer to the research question

State the measured relationship or result, cite the processed evidence that supports it and say whether the stated hypothesis is supported, contradicted or remains unresolved. Do not repeat the procedure.

Example — evaluating an inverse model

If a hypothesis predicts $y\propto1/x$, values of $xy$ that clearly vary beyond their uncertainties contradict that prediction. Values that are approximately constant are consistent with the model, but they do not prove it: another relationship may also fit the limited range.

Uncertainty controls confidence, not truth

Agreement within stated uncertainty supports consistency with an accepted value; disagreement outside it flags tension, underestimated uncertainty or systematic effects. Neither result alone proves the model or identifies the cause.

Evaluate by tracing cause → measurement → result → conclusion

Name a specific random or systematic effect, explain which measurement it changes and in what direction when known, then state how the processed result and conclusion are affected. A label such as “human error” is not an evaluation.

Evaluate the hypothesis, not just the apparatus

Use trend, intercept, uncertainty bars and alternative explanations to state whether the evidence supports or contradicts the hypothesis. A best-fit line missing some uncertainty bars weakens the proposed model; an approximately constant transformed quantity is compatible with a model but cannot prove it.

Make improvements specific and realistic

For each weakness, name the physical change, how it reduces or measures the stated effect and why the conclusion becomes stronger. “Be more careful” and unrelated extra repeats do not correct a systematic bias or an unrealistic assumption.