Designing a Biological Experiment: Units, Controls and Randomisation · seed 1 · A4, ink-friendly. The answer key prints on its own page for grown-ups.

Designing an experiment that can convince

Science · Scientific Inquiry · ages 22-23
Name ______________________   Date ____________

  1. What defines an experimental unit?

    • The smallest entity assigned independently to treatment
    • Every reading recorded separately by a measuring device
    • The entire collection of entities measured in the study

  2. Which information belongs in the design before study data are collected?

    • The final effect, its p-value and the accepted conclusion
    • The treatments, experimental units and response to measure
    • The observed group means, final plots and publication date

  3. Random assignment guarantees that every known and unknown baseline factor is balanced in the realised treatment groups.

    Circle one:   True   False

  4. Patients attend clinics with different baseline outcomes. Each clinic can supply patients to both treatments. Which design uses clinic as a block?

    • Group patients by clinic and randomise within each clinic
    • Give one clinic the new treatment and another the control
    • Pool all clinics and assign treatment by arrival order

  5. One tray containing 24 plants is independently assigned to a dose. How many treatment units does that dose have?

    • 24 units, counting every plant measured in the tray
    • 25 units, counting both the tray and all its plants
    • 1 unit, counting the independently assigned tray

  6. A trial compares an active pill with a matched placebo pill under the same care schedule. What comparison is the placebo intended to support?

    • A comparison of active doses under the same care schedule
    • A comparison with a similar pill lacking the active drug
    • A comparison of the care schedule with no pill supplied

  7. A trial randomises treatments among volunteers from one gym. Which limitation remains when applying its result to all adults?

    • Random assignment makes gym volunteers representative of all adults
    • Gym recruitment mainly limits treatment randomisation within the study
    • Gym volunteers may not represent the wider adult population

  8. Two cages each contain six mice. One whole cage receives treatment and the other control. The report counts 12 treatment units. What is the design problem?

    • Animals should be the units because each has a measured response
    • Taking a mean per cage supplies enough treatment replication
    • Cages are the units, with only one cage per treatment
LightMySky · lightmysky.comW1-mt_Hc6BAdlkLf-s1

Answer key

For grown-ups. Fold this page away before handing over the rest.

Designing an experiment that can convince W1-mt_Hc6BAdlkLf-s1

  1. The smallest entity assigned independently to treatment · The assignment defines the unit. Separate readings do not necessarily receive separate treatments.
  2. The treatments, experimental units and response to measure · A design states what receives treatment and what will be measured. Results are not yet available.
  3. False · Random assignment balances factors in expectation. Chance imbalances can remain in a particular sample.
  4. Group patients by clinic and randomise within each clinic · Within-clinic randomisation permits treatment comparisons within each clinic rather than mixing treatment with clinic.
  5. 1 unit, counting the independently assigned tray · The dose was assigned to one tray. The plants provide measurements within that treatment unit.
  6. A comparison with a similar pill lacking the active drug · The matched placebo retains a similar pill procedure without the active drug. It is not another active dose or a no-pill comparison.
  7. Gym volunteers may not represent the wider adult population · Random treatment assignment does not turn a narrow volunteer sample into a representative sample of all adults.
  8. Cages are the units, with only one cage per treatment · There are two assigned cages in total, one per treatment. Cage differences cannot be separated from treatment differences in this design.
Worksheet · LightMySky