
use peer critique and a test run on sample data to improve the trial plan, judge whether its evidence would be sufficient, and defend its conclusions against counter-claims.
Phase 4: Build
Do Now. Look at your version 1 checklist. Write one criterion you are sure you met and one you are not, and say why.
A plan puts all of Lot A along the sunny south edge of each bed and Lot B in the middle. What is the flaw?
Gallery critique. Teams set out their bed maps, matrices and statistics plans. Visitors leave sticky notes in three colors: I like… (a strength backed by evidence), I wonder… (a question about the design or statistics), and What if… (a specific change). Critique the design, not the designer.
Class criteria (set in 4.07). A garden trial plan for the seed library must (1) separate genetic from environmental effects on a trait, (2) grow at least 30 plants per group, (3) place plants at random within each bed, (4) fit in four raised beds of 1.2 m × 2.4 m, (5) finish within one growing season, and (6) cost no more than 12 sample budget points.
Worked example 1: a test run on sample data. Mei’s team ran its version 1 plan on the sample trial data from 4.07. Their summary said, “Lot A was bigger, so big seeds have better genes.” A critic wrote: I wonder… Lot A beat Lot B by only 0.2 mm in each bed, and both SDs are 0.5–0.7 mm. Is that difference real? The team rechecked: a difference smaller than the spread within groups is weak evidence. Their revised conclusion: the trial shows a strong environmental effect (about 1.4 mm) and no clear genetic effect of saved-seed size.
Worked example 2: version 2. Critique also caught that the team’s beds were along a fence that shaded one end. Version 2 added: random placement within each bed, a second season’s repeat written in as optional, and a rule to record any plant lost to pests rather than replacing it. Each change makes the evidence more reliable.
Is the evidence sufficient? Judge a trial’s evidence with three checks. Enough data: at least 30 plants per group and, ideally, a second season. Valid comparison: only one factor differs in each comparison, and placement is random. Cause, not just correlation: the trial changed a factor on purpose, so a difference is evidence of cause, unlike the returned-seed table, which only showed an association.
Defending against counter-claims. A volunteer says, “Big beans still have better genes; your trial was just unlucky.” A strong reply states the claim, names the evidence (two beds, 30 plants per group, random placement, a 0.2 mm lot difference smaller than the within-group spread) and explains why it outweighs the counter-claim, while admitting what one season cannot rule out.
| Version | Separates genes and environment? | Plants per group | Random placement? | Cost (points) | Criteria met (1-6) |
|---|---|---|---|---|---|
| 1 | |||||
| 2 |
Use your critique notes and both versions of the plan.
Evaluate and defend your plan.
Judge your version 2 plan: is the evidence it would produce sufficient, valid and reliable? Name one thing it still cannot rule out. Then answer the counter-claim “big beans have better genes” using the sample trial data.
In the 1930s and 1940s, botanists Jens Clausen, David Keck and William Hiesey ran a famous set of experiments for the Carnegie Institution. They collected wild plants, such as yarrow, from populations living from the coast to high mountains. They then split single plants into pieces, so every piece was a genetically identical clone, and planted the pieces in gardens at three elevations, from near sea level to high in the mountains. Clones of the same plant grew to different heights in different gardens: evidence of the environment’s effect. But plants from different home populations still differed when grown side by side in the same garden: evidence of genetic differences. By using clones and common gardens, they separated genes from environment, the same logic as the seed library’s trial.
Think about it: Which feature of the experiment held genes constant, and which held the environment constant? How does your trial plan do the same?
Trial plan version 2 feeds straight into the variation report.
Exit ticket: which change from version 1 to version 2 most improved the trial's evidence, and why?
How confident are you that you can improve a trial plan with critique and defend its conclusions against counter-claims.
Materials: version 1 plans, sticky notes in three colors, the sample trial data, calculators, revision-log template.
Timing: about 15 minutes of gallery critique, 20 minutes to test on sample data and revise, 15 minutes for the case study, career and log.
Listen for: small differences treated as real; “randomizing is unfair”; rebuttals that attack the person rather than the evidence.
Support / stretch: give a counter-claim response frame. Stretch: compute the standard error of each group mean (SD ÷ √30) and decide whether the 0.2 mm lot difference is larger than about two standard errors.