Skip to content

Experimental Design Anatomy

Run the Experimental Design Anatomy MicroSim Fullscreen

You can include this MicroSim on your website using the following iframe:

1
2
<iframe src="https://dmccreary.github.io/fft-benchmarking/sims/experimental-design-anatomy/main.html"
        height="437px" width="100%" scrolling="no"></iframe>

About This MicroSim

A well-formed benchmark question already contains its own experimental design. You do not invent the variables afterward — if the question is written properly, they are sitting in the sentence, and this simulation is the act of pulling them out.

Every phrase in the example plays exactly one of four roles:

  • Research question — what the whole experiment exists to answer. Not a variable, and not measurable on its own.
  • Independent variable — the one thing you deliberately change between runs. One. If two things change, no result can tell you which one mattered.
  • Dependent variable — the number you read off at the end. It must be a number, with a unit, or you cannot compare two runs.
  • Controlled variables — everything held identical across every run. Usually the longest list, and the one people forget to write down.

The pattern is easiest to see in the sentence structure itself. "Does A change B, holding C, D, and E fixed?" — A is what you vary, B is what you measure, and C through E are what you pin down. Once you can hear that shape, you can write it, and once you can write it you have a benchmark suite: one run per value of A, recording B, with C through E identical every time.

The three examples come from different domains — audio tone detection, bearing vibration, and radio signal separation — because the structure is the point, not the subject. The same four boxes work for all of them.

A wrong placement gets you a hint about the category, not the answer. If you drop "peak-frequency error in Hz" into Independent Variable, the simulation tells you that phrase describes something you measure rather than something you change, and leaves you to work out where it goes.

How to Use

  1. Read the whole question first. The highlighted phrases are the ones that have a role to fill.
  2. Drag each phrase into the box where you think it belongs. A dashed outline marks where it came from in the sentence.
  3. Press Check my answers. Correct phrases turn green; the first incorrect one turns red and produces a hint.
  4. Fix and re-check until all six are green.
  5. Press Try a new example for a question from a different domain, and see whether the structure you just learned transfers.

Lesson Plan

Grade Level: Undergraduate

Duration: 10-12 minutes

Prerequisites:

  • The idea that an experiment compares outcomes under changed conditions
  • Familiarity with the FFT parameters used in the examples

Learning Objective: Classify the components of a stated research question into research question, independent variable, dependent variable, and controlled variables, and construct a benchmark suite around them.

Activities:

  1. Classify the first example (3 min). Work through the audio example individually. Most errors are between independent and dependent — the hint text addresses that pair directly.
  2. Transfer to a new domain (3 min). Do the vibration example without discussion. Students who have understood the structure finish it much faster than the first.
  3. Construct the suite (3 min). With all six phrases placed, ask: how many benchmark runs does this question require, and what does each row of your results table hold? (Two runs — one per value of the independent variable — each recording the dependent variable, with the controlled variables recorded once in the header.)
  4. Break the design (3 min). Ask what happens if a second independent variable sneaks in — say the clock speed also changes between runs. Establish that the result becomes uninterpretable, not merely noisy.

Assessment: Write a research question, in the same "Does A change B, holding C fixed?" form, for a benchmark you would actually want to run on the Pico 2. Then list the runs your suite would contain.

References