All outcomes
Skills

Biology Experimental Design

6 weeks · 0 milestones

Design a real biological investigation from scratch: state a testable hypothesis, identify independent and dependent variables, define controls, justify sample size using a power analysis or reasoning from published literature, and document ethics considerations for living organisms or human participants. The proof is the complete experimental design document reviewed and approved by a working biologist BEFORE data collection begins — the biologist confirms that the design is methodologically sound and the hypothesis is genuinely testable. For students without lab access, the design may be for a secondary data analysis study using a named publicly available dataset (GBIF, NCBI SRA, iNaturalist) — the design document requirements are identical: hypothesis, variables, sampling methodology, statistical analysis plan. The design precedes data collection; this is a genuine prerequisite, not a formality.

Milestone map

Milestone map

3 milestones

Formulate hypothesis and write protocol

1–2 weeks

Define a testable biological hypothesis and design the experiment that will test it. Accessible alternative: computational experiments (simulations, mathematical models, re-analysis of a published raw dataset using a novel design approach) are fully valid for this outcome — field or wet-lab access is not required. The proof standard is identical for both routes.

Proof required

Submit a written experimental protocol (2–4 pages) including: your null and alternative hypotheses, your independent and dependent variables, all controlled variables, your replication strategy, a brief statistical power justification (why your chosen sample size is adequate), and the analysis you plan to run on the results.

What gets checked

  • Null hypothesis is falsifiable and clearly stated — not 'I expect to find a difference' but 'there is no significant difference in [variable] between [condition A] and [condition B]'
  • All three variable types are named (independent, dependent, controlled) with specific operationalised definitions
  • Statistical power justification is present — even a brief note ('n=10 per group based on a published effect size of d=0.8 at 80% power') demonstrates the submitter understands why sample size matters

Common mistakes

  • Writing a protocol without thinking through controls — common error is designing an experiment where confounders cannot be distinguished from the treatment effect
  • Choosing a sample size without justification — post-hoc power issues account for a large proportion of non-replicable biology findings; the protocol must address this before data collection

Resources

Foundationstart here

Depthgo deeper

What a verifier looks for

  • Ask the submitter to state their null hypothesis in one sentence — it should be specific enough to be clearly falsified by a result.
  • Ask how they chose their sample size — they should be able to cite a prior effect size estimate or justify the sample size through power logic.
  • Ask what would change about the interpretation if their main confound were not controlled — a well-designed experiment has a ready answer to this.

Execute experiment and document results

2–6 weeks depending on organism or simulation

Run the experiment according to your protocol and collect raw data. Document all deviations from the protocol as they occur — a lab notebook entry that notes problems is more credible than a clean log that pretends nothing went wrong. For computational experiments: run the model or re-analysis, log each parameter decision, and record all intermediate outputs.

Proof required

Submit your raw data table or computational output files, your protocol execution log documenting what actually happened (including any deviations from plan), and a brief preliminary analysis showing summary statistics or an initial figure for the main measured variable.

What gets checked

  • Raw data is structured consistently — columns labelled, units stated, each row is a single observation
  • Execution log records the actual dates and conditions of data collection — not a post-hoc reconstruction
  • At least one deviation from protocol is noted — a log with zero deviations is rare in practice and reduces credibility

Common mistakes

  • Not logging deviations as they occur and instead writing a 'clean' version after the fact — reviewers can detect post-hoc reconstruction in a Q&A
  • Not performing the planned analysis until M3 — running at least the summary statistics in M2 allows early detection of data quality problems while it is still possible to address them

Resources

Foundationstart here

What a verifier looks for

  • Ask the submitter to describe a specific deviation from their protocol and how they decided to handle it — genuine experimenters always have one.
  • Ask how they verified data entry accuracy — transcription errors in raw data tables are extremely common and a good experimenter has a checking strategy.
  • Ask what the preliminary figures suggested before running the final analysis — confirms they engaged analytically with M2, not just data collection.

Write report and present with Q&A

1–2 weeks to write and schedule review

Complete a full research report and present to a biologist who will challenge your experimental design choices. The Q&A session tests genuine scientific reasoning — the reviewer will raise alternative explanations, design limitations, and 'what if' scenarios that require real biological understanding to address.

Proof required

Submit your final research report (3–5 pages: hypothesis, methods, results, discussion, limitations) and a written Q&A record showing the reviewer's specific challenges and your responses. The reviewer must be named and their biological research background stated.

What gets checked

  • Report includes a limitations section that addresses the specific design choices made in M1 — not a generic 'sample size could be larger'
  • Q&A record shows at least two design-specific challenges from the reviewer (not confirmation of results) and substantive responses
  • Discussion section engages with at least one alternative explanation for the results, not just the intended interpretation

Common mistakes

  • A discussion that only confirms the hypothesis without engaging with what alternative mechanisms could produce the same data — this is the most common scientific writing failure mode at undergraduate level
  • Recruiting a reviewer who only asks about the results, not the design — the Q&A must include at least one challenge about a methodological choice made in M1

Resources

Foundationstart here

What a verifier looks for

  • Ask what the reviewer's strongest methodological challenge was and how the submitter answered it — someone who went through a genuine Q&A can recall this clearly.
  • Ask what the main alternative explanation for the results is and how the design did or did not rule it out — a genuine experimenter has thought about this.
  • Verify the reviewer has biological research experience (not just biology teaching) — they need to be able to generate design-specific challenges, not just read comprehension questions.

We use analytics to improve Powstik. No ads, ever.