Design an experiment that can answer its question

The question decides what to measure

Suppose two paper-airplane folds look different and you want to know which travels farther. “Which is best?” is underspecified: best could mean greatest distance, longest time in the air, easiest construction, or most consistent performance. Choose one outcome before comparing results.

For this investigation, define distance as the straight-line distance from the launch mark to the point where the plane first touches the floor. This operational definition prevents one observer from measuring the final resting point while another measures the first impact. Write the definition beside the data table so that it travels with the numbers.

The experiment is a low-risk classroom activity only when there is adequate clear space and adult permission. Do not launch toward people, use sharp materials, or interfere with normal access through a room.

Identify the change and competing explanations

The independent variable is the fold design. The dependent variable is measured flight distance. Try to keep paper size, paper type, builder instructions, starting point, and launch method comparable. If the second design also uses heavier paper, a difference cannot be attributed specifically to the fold.

Some conditions cannot be held perfectly constant. A human launcher varies force and angle, and a small indoor air movement may affect a flight. Acknowledge these factors and reduce them where practical. A consistent instruction such as “release from the marked height using the same grip” is more informative than an unexplained statement that conditions were identical.

Alternate the order of designs across trials instead of completing every flight of A before B. This can reduce confounding with a change in launcher fatigue or room conditions. It does not remove all variability.

An illustrative dataset

The following numbers are invented for teaching. They are not results from an actual school experiment.

A has mean 5.0 meters and range 0.4 meters. B has mean 5.0 meters and range 3.8 meters. The means are equal, but the consistency is very different. If the goal is reliably reaching a target distance, that distinction matters.

The range uses only the extreme values, so it is a simple description rather than a complete account of spread. Keep the raw observations and display them; a summary should make a dataset easier to understand, not make it disappear.

Do not erase inconvenient observations

An unusual flight may reflect natural variability or a procedural error. Those are different explanations. If a person accidentally launches from the wrong mark, document that departure and decide how to handle it using a rule established before looking at which design benefits.

If a flight is merely shorter than expected, deleting it can distort the comparison. Report the complete dataset and investigate the discrepancy. A transparent note such as “trial 4 used the wrong starting line and was repeated under the stated procedure” is better than silently replacing a number.

Measurement precision also matters. A ruler marked in centimeters does not justify reporting every distance to many decimal places. Record a level of precision supported by the instrument and method.

Write a conclusion proportional to the investigation

A defensible conclusion for the teaching dataset is: “The two designs had the same average measured distance in these five trials, while A’s distances varied less.” It describes both the measured outcome and the scope.

“A is always better” adds a criterion and extends the result to situations not tested. “B is defective” interprets variability without evidence about its cause. An explanation should connect the claim with the actual numbers, then identify limits such as small trial count, one launcher, and one room.

If you suspect that the fold is more sensitive to release angle, design a follow-up that changes angle systematically while keeping other conditions comparable. This turns uncertainty into another testable question.

Use the trial analyzer as a record, not a verdict

The analyzer calculates a count, mean, median, and range from the measurements you enter. Its dot plot shows individual values along a number line. It does not determine whether the investigation is fair, whether observations were recorded correctly, or whether a difference is statistically meaningful.

Enter the two datasets separately and compare their plots. Then create a dataset with the same mean but a different range. Explain why a conclusion based only on the mean would miss an important feature. The lesson is about the relationship between a question, a method, and a record—not about selecting the largest number as a winner.

Your lab notebook page

1. State the question and define the outcome precisely.
2. Draw or describe the designs so someone else can reproduce them.
3. Record controlled conditions and factors that may still vary.
4. Write every observation with units and relevant procedural notes.
5. Show calculations and a plot, then state a bounded conclusion.
6. Propose one follow-up that addresses a specific limitation.

For background on observation, hypotheses, and investigation, see OpenStax’s introduction to the science of biology. Its broader discussion complements this original classroom example.